EP4366518A1 - A method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the method - Google Patents
A method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the methodInfo
- Publication number
- EP4366518A1 EP4366518A1 EP22757642.8A EP22757642A EP4366518A1 EP 4366518 A1 EP4366518 A1 EP 4366518A1 EP 22757642 A EP22757642 A EP 22757642A EP 4366518 A1 EP4366518 A1 EP 4366518A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- horse
- stall
- frequency
- detected
- processing device
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 241000283073 Equus caballus Species 0.000 title claims abstract description 193
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000012545 processing Methods 0.000 claims abstract description 46
- 230000008859 change Effects 0.000 claims abstract description 38
- 230000000007 visual effect Effects 0.000 claims abstract description 38
- 230000000694 effects Effects 0.000 claims abstract description 23
- 238000013473 artificial intelligence Methods 0.000 claims abstract description 21
- 238000013528 artificial neural network Methods 0.000 claims abstract description 14
- 238000010191 image analysis Methods 0.000 claims abstract description 10
- 238000004458 analytical method Methods 0.000 claims description 21
- 238000004891 communication Methods 0.000 claims description 12
- 210000003608 fece Anatomy 0.000 claims description 11
- 239000010871 livestock manure Substances 0.000 claims description 11
- 230000027939 micturition Effects 0.000 claims description 11
- 230000009471 action Effects 0.000 claims description 10
- 210000001364 upper extremity Anatomy 0.000 claims description 10
- 230000004913 activation Effects 0.000 claims description 9
- 239000003814 drug Substances 0.000 claims description 9
- 229940079593 drug Drugs 0.000 claims description 9
- 239000013589 supplement Substances 0.000 claims description 8
- 210000001015 abdomen Anatomy 0.000 claims description 7
- 208000004998 Abdominal Pain Diseases 0.000 claims description 6
- 208000002881 Colic Diseases 0.000 claims description 6
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 6
- 238000005096 rolling process Methods 0.000 claims description 6
- 206010013975 Dyspnoeas Diseases 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 230000003213 activating effect Effects 0.000 claims description 5
- 208000035475 disorder Diseases 0.000 claims description 5
- 230000035622 drinking Effects 0.000 claims description 5
- 210000003141 lower extremity Anatomy 0.000 claims description 5
- 238000009423 ventilation Methods 0.000 claims description 5
- 241001465754 Metazoa Species 0.000 description 24
- 230000006399 behavior Effects 0.000 description 20
- 238000005259 measurement Methods 0.000 description 11
- 238000001514 detection method Methods 0.000 description 10
- 241000283086 Equidae Species 0.000 description 9
- 230000036541 health Effects 0.000 description 8
- 238000013459 approach Methods 0.000 description 7
- 241000283690 Bos taurus Species 0.000 description 6
- 230000007246 mechanism Effects 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 5
- 230000035935 pregnancy Effects 0.000 description 5
- 230000003187 abdominal effect Effects 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 4
- 230000007423 decrease Effects 0.000 description 4
- 230000029058 respiratory gaseous exchange Effects 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 206010037660 Pyrexia Diseases 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 125000004122 cyclic group Chemical group 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 206010012735 Diarrhoea Diseases 0.000 description 2
- 206010061218 Inflammation Diseases 0.000 description 2
- 244000250129 Trigonella foenum graecum Species 0.000 description 2
- 210000000988 bone and bone Anatomy 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 2
- 210000003414 extremity Anatomy 0.000 description 2
- 230000005802 health problem Effects 0.000 description 2
- 230000004054 inflammatory process Effects 0.000 description 2
- 230000000241 respiratory effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 241000894007 species Species 0.000 description 2
- 208000011623 Obstructive Lung disease Diseases 0.000 description 1
- 238000004378 air conditioning Methods 0.000 description 1
- 230000000740 bleeding effect Effects 0.000 description 1
- 230000036760 body temperature Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001055 chewing effect Effects 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 210000002249 digestive system Anatomy 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000005713 exacerbation Effects 0.000 description 1
- 230000037406 food intake Effects 0.000 description 1
- 235000012631 food intake Nutrition 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 208000030175 lameness Diseases 0.000 description 1
- 244000144972 livestock Species 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 208000004396 mastitis Diseases 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 210000004197 pelvis Anatomy 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000036387 respiratory rate Effects 0.000 description 1
- 208000023504 respiratory system disease Diseases 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
- 238000002076 thermal analysis method Methods 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 210000001635 urinary tract Anatomy 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01K—ANIMAL HUSBANDRY; AVICULTURE; APICULTURE; PISCICULTURE; FISHING; REARING OR BREEDING ANIMALS, NOT OTHERWISE PROVIDED FOR; NEW BREEDS OF ANIMALS
- A01K29/00—Other apparatus for animal husbandry
- A01K29/005—Monitoring or measuring activity, e.g. detecting heat or mating
Definitions
- the invention relates to a method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the method.
- Japanese patent application JP2020014421 A relates to a livestock birth prediction system using a image acquisition unit for acquiring image of a pre-partum animal, a monitoring unit for monitoring animal's behaviour based on an image from the image acquisition unit and a predictive unit for predicting the delivery date based on animal's behaviour.
- Analysis of behaviour is a dynamic analysis and is conducted based on frames obtained at specific time intervals. The obtained results are compared with the previous behaviour pattern of a given animal. Parameters such as duration of activity, standing or eating by the animal are taken into account. The system takes into account changes in these behaviours.
- the system is also provided with means for notifying the animal’s handler of the expected moment of delivery.
- the international patent application WO2021237144A1 describes a system equipped with a camera which enables determining health condition of an animal based on an image from the camera. This image is used by a neural network which compares it with models available for a given animal and other animals of the same species, and allows to determine the health condition of a given animal. Camera images are images showing depth, preferably from a near-IR camera, but the use of a conventional optical camera has also been mentioned.
- a camera is also used to determine health condition of a subject based on the calculations of the processor which compares the obtained images with the data available in the database and determines the health of an individual (using models, artificial intelligence and neural networks).
- the device can then generate information about the health condition - e.g. mastitis, fever, etc.
- the device is based on parameters related to temperature, but also on the observation of given features of an individual (e.g. an animal, including a horse) or its activity - heartbeats (additional measuring device), number of breaths (additional measuring device), movements.
- the device also transmits alarm messages.
- European patent EP3337422B1 describes prediction of calving moment based on images from a 3D camera.
- the image from the camera is processed and the parameters characteristic for the calving are determined, and then its time is predicted.
- the parameters determined include: body contractions, position (orientation) of a cow, distance between pin bones (bones within cow's pelvis), lying down of a cow, and tail movements.
- the obtained data is compared with the data in the database to determine the moment of calving.
- the system is also equipped with an alarm system to alert about upcoming delivery.
- the aim of the invention is to develop a solution that would enable accurate and precise, and at the same time versatile, detection and prediction of a number of particular conditions of a horse, including in particular foaling of a mare, using only images from a visual camera for measurements.
- the aim of the invention was to provide a system that would allow to obtain an automatic response of external devices to the detected situation, which would contribute to increasing the safety of horses while limiting the involvement of horse’s handler.
- the subject of the invention is a method for predicting and detecting particular conditions of a horse, in particular for predicting the moment of foaling.
- a processing device having: an input where an image of said horse is received from at least one visual camera, an output which can assume at least two different output states, and an artificial intelligence model, such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data, is used.
- the change of the output state of the processing device causes activation of at least one accessory device selected from: additional camera, lighting, alarm device, actuator, electric lock and communication module.
- step a) additionally at least one parameter selected from: the frequency of passing of manure, the shape of the stool, the frequency of urination, pawing with front leg or legs, the frequency of horse looking at its sides, the frequency of kicking the hind legs of the horse up towards the abdomen, the frequency of rolling, the frequency of eating and the frequency of drinking, is determined.
- the image from the visual camera is further analysed for the occurrence of abnormal horse body positions, such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
- abnormal horse body positions such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
- the change of the output state of the processing device results in activation of an alarm device.
- the change of the output state of the processing device results in turning on a light at the location of the horse or activating at least one additional camera in the horse’s stall.
- one or more of the following actions is initiated: a. if the horse’s foaling is predicted or detected, the surface area of the horse's stall is automatically increased or an extra portion of bedding is provided to additionally make the horse’s stall, b. if dyspnoea is predicted or detected in the horse, the stall is automatically opened to allow the horse to leave the stall, the ventilation system at the location of the horse is activated, or drugs or supplements are automatically dosed, c. if colic, fewer or a disorder of passing of manure or urination in the horse is predicted or detected, drugs or supplements are automatically dosed.
- the subject of the invention is also a system for predicting and detecting particular conditions of a horse, in particular for predicting the moment of foaling.
- the system comprises at least one visual camera and a processing device having:
- an artificial intelligence model such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data.
- the system components are configured and programmed to: a) determine the following parameters by means of image analysis:
- the system further comprises one or more accessory devices selected from: additional camera, lighting, alarm device, actuator, electric lock and communication module, configured and programmed such that the change of the output state of the processing device results in activation of one or more of said accessory devices.
- accessory devices selected from: additional camera, lighting, alarm device, actuator, electric lock and communication module, configured and programmed such that the change of the output state of the processing device results in activation of one or more of said accessory devices.
- the system is additionally configured and programmed to determine at least one parameter selected from: the frequency of passing of manure, the shape of the stool, the frequency of urination, pawing with front leg or legs, the frequency of horse looking at its sides, the frequency of kicking the hind legs of the horse up towards the abdomen, the frequency of rolling, the frequency of eating and the frequency of drinking.
- the system is additionally configured and programmed to analyse the image from the visual camera for the occurrence of abnormal horse body positions, such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
- abnormal horse body positions such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
- the system is additionally equipped with an alarm device, and is configured and programmed such that the change of the output state of the processing device results in activation of the alarm device.
- the system is additionally configured and programmed such that the change of the output state of the processing device results in turning on a light at the location of the horse or activating at least one additional camera in the horse’s stall.
- the system is additionally configured and programmed such that, depending on the specific detected particular condition of the horse, one or more of the following actions is further initiated: a. if the horse’s foaling is predicted or detected, the surface area of the horse's stall is automatically increased or an extra portion of bedding is provided to additionally make the horse’s stall, b. if dyspnoea is predicted or detected in the horse, the stall is automatically opened to allow the horse to leave the stall, the ventilation system at the location of the horse is activated, or drugs or supplements are automatically dosed, c. if colic, fewer or a disorder of passing of manure or urination in the horse is predicted or detected, drugs or supplements are automatically dosed.
- the subject of the invention is also the above described system, configured and programmed to carry out the above described method.
- Fig. 1 shows schematically an implementation of the method of invention in one embodiment, in which the output state triggers the activation of a communication module, such as for example a Wi-Fi module.
- a communication module such as for example a Wi-Fi module.
- Fig. 2 shows schematically additional functionality of the method in which the communication module may be a module providing a two-way communication between horse’s handler and the processing device in order to obtain data about the horse from the processing device or at least one camera via user interface.
- the communication module may be a module providing a two-way communication between horse’s handler and the processing device in order to obtain data about the horse from the processing device or at least one camera via user interface.
- Fig. 3 shows an exemplary implementation of a user interface.
- Fig. 4 shows an exemplary embodiment of measuring the number of breaths of a horse according to the invention.
- Fig. 5 shows another embodiment of measuring the number of breaths of a horse according to the invention.
- Fig. 6 shows the results of measuring the number of breaths of a horse according to the invention.
- the method and system of the present invention for predicting and detecting particular conditions of a horse provide the possibility of a quick, automatic response of the system to the detection of a life-threatening or health-endangering condition of an animal, or a condition that could potentially have serious consequences for the horse or might require veterinary intervention.
- the possibility of producing various output states at the output of the system provides the possibility of activating a suitable device depending on a specific, detected condition, thus minimizing the need for often time-consuming interference of the horse’s handler.
- This substantially maintenance- free system increases safety of the horse, which is provided with quick, often automatic help, and at the same time increases safety of the user by minimizing the need for human to be in the presence of animal, which, especially in health- endangering or life-threatening situations, may behave in a dangerous manner.
- the system does not eliminate human being - horse's handler - from a series of taken actions.
- an alert informing about detection of such condition is immediately sent to the horse’s handler. This allows for a quick human intervention, if needed.
- the method and system can be used to detect both the upcoming delivery in mares and to predict with high accuracy, even up to approximately 90%, when such delivery will take place (both short-term and long-term), as well as to detect conditions such as: problems with urination or passing of manure, disorders connected with passing of manure, colic conditions, and even respiratory disorders, especially important in horses with RAO (recurrent obstructive pulmonary disease), and even fever.
- RAO recurrent obstructive pulmonary disease
- a particular advantage of the present invention is the fact that for the correct and reliable operation of the method and system of the invention, a camera providing only image data, i.e. a visual camera, is required.
- This feature allows to easily implement the method of the invention in facilities already equipped with monitoring, such as industrial monitoring, without the need to install accessory equipment for collecting additional information about the horse.
- the method of the invention is therefore also completely non-invasive for the horse.
- the invention relates to a method for predicting and detecting particular conditions of a horse.
- a particular condition of a horse is defined as a condition of a horse that deviates from the normal behaviour or health condition of a particular individual. It should be realized that a given behaviour which is an indication of a particular condition for one individual may not be an indication of a particular condition for another individual, which exhibits given type of behaviour with a substantially constant frequency.
- a given behaviour which is an indication of a particular condition for one individual may not be an indication of a particular condition for another individual, which exhibits given type of behaviour with a substantially constant frequency.
- An example is the lying down of horses - some of them lie down very often and/or for a long time, while others practically do not lie down at all. Exemplary particular conditions detectable or predictable by the method of the present invention will be described below in this document in details.
- a processing device is to be understood as any device that enables processing of an image from a visual camera in order to analyse and measure at least one of selected features.
- the processing device has an input (by means of which image data from at least one visual camera connected thereto is provided) and an output.
- a processing device is placed at the location of the horse, such as, for example, stall, stable, standing stall, run pen, hospital, corral (the use of one of these terms herein does not exclude implementation of the described feature with respect to other, even not directly mentioned locations if such an implementation is physically possible), and is connected by a wire or wirelessly to the at least one visual camera.
- a visual camera is generally defined as a camera that collects image-only data, however, in alternative embodiments, other types of cameras, such as audio-visual cameras, may be used.
- the output from the processing device is adapted to generate at least two output states and is connected to at least one external device adapted to receive the output state and to perform a particular action depending on the particular received at least one output state.
- an output state is to be understood as a logical state such as true/false or 0/1 , a change in the position of a component of the processing device, a change in the voltage at the output of the processing device, the appearance or disappearance of a signal at the output of the processing device - in particular such as electrical, radio or optical signal - the output state can be a specific physical value - electrical (voltage, current, impedance, capacity), radio or optical.
- the output state may be communicated to the corresponding external device via an interpreter which interprets the output state or change of the output state and communicates the interpreted state to the corresponding external device.
- each of the external devices may be equipped with an individual interpreter.
- Examples of external devices that may be used in the present invention are: an additional camera, lighting, alarm device, actuator, electric lock and communication module.
- the communication module may be, for example, a GSM module, a Wi-Fi module, a Bluetooth module and other such modules.
- the actuator may be, for example, a power switch for air conditioning or ventilation, a power switch for a drug dispenser, and any other suitable device for carrying out a given physical action. Exemplary implementations of actuators will be described hereinbelow.
- a user terminal such as a mobile phone, tablet or computer
- the processing device can monitor condition of an animal.
- each of the stalls may be equipped with a suitable device enabling the user to review parameters relating to the condition of an animal.
- the processing device is furthermore equipped with an artificial intelligence model which is designed to analyse an image from at least one visual camera and interpret it in order to detect the particular condition of a horse.
- the artificial intelligence model are neural networks, even more preferably properly trained neural networks, however, it should be understood that any artificial intelligence model may be used. Random forests are an alternative example of an artificial intelligence model.
- the basis of the method of the invention is to use an artificial intelligence model and to teach it the behaviour of a specific horse in order to detect abnormalities in animal’s behaviour which may be indicative of an occurrence of a particular condition.
- the image from the at least one visual camera is analysed in a first step (step a)) of the method in order to measure the following parameters:
- the specific parameters related to the occurrence of abnormal horse body positions such as caused by the horse's leg being stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse, may be additionally analysed.
- step b) of the method the obtained parameter values are analysed by means of an artificial intelligence model in order to detect condition of a horse, and then in step c) it is determined whether the detected condition of the horse is a particular condition by comparing the obtained parameter values with at least one particular condition of a horse. Due to the fact that the evaluation of condition of a horse takes into account, among other things, changes in the frequency of occurrence of behaviours, it should be taken into account that the change may be caused by change of season, weather, temperature of the environment and a number of other parameters, and it should be interpreted with a certain margin of permissible fluctuations.
- the foaling is intended to describe not only the duration of the birthing action, but also the period of preparations of the horse's body for puerperium and the period of pregnancy, during which, in the image analysis, any changes in the course of pregnancy are noticeable.
- the method of the invention also makes it possible to observe and analyse vital parameters of the new-born offspring during the first hours or even weeks of their life. Such measurements may contribute to the detection of early abnormalities of neonatal and adolescent age of foals, which will enable quick detection of e.g.
- the procedure for detection of the parameters based on an image from at least one visual camera will be exemplified on the basis of analysis of number of breaths of a horse, also referred to as respiratory rate and typically given as number of breaths per minute.
- ROI region of interest
- the procedure for analysing activity comprises the following steps:
- a segmentation-type algorithm is run for each frame of the video material from the visual camera.
- the level of changes between frames is determined. • The value is scaled to a specified activity scale, averaged over a series of frames, and returned as the activity level.
- a different algorithm which analyses horse's position for each frame can also be used for this purpose.
- the analysis may be based on determining contour of the horse and observing changes of the contour in individual frames or on determining the position of individual parts of the horse's body (e.g. limbs) and examining their movement in subsequent frames.
- Horse's halter catching on elements of a stall or stall’s equipment e.g. stall’s rungs, drinker, manger etc.
- Horse's leg being stuck in the elements of a stall or stall’s equipment e.g. hay net, manger, stall’s rungs etc.
- the last of the mandatory parameters to be determined serves to determine whether the animal is standing or lying down at the moment.
- a binary classifier was used.
- the implemented algorithm detects the horse's position in a given frame of the video material and classifies it as a standing / lying down position, respectively.
- At least one change of the output state is made at the output of the processing device.
- the change of the output state is communicated to an external device, which activates a programmed action of the external device - each at least one change of the output state is associated with a given, specific action of at least one external device. Exemplary actions for exemplary external devices will be presented below:
- the additional camera is to be defined the same way as the visual camera defined hereinabove.
- it may be an audio visual camera or any other camera, preferably located at a different angle and/or height than the primary camera, which allows to view the animal from different directions.
- a camera is a thermal imaging camera, which additionally enables the estimation of the animal's body temperature (for example: temperature increase - disease, temperature drop as low as the ambient temperature - death), as well as the observation of local inflammations (e.g. inflammation of the limb, indicating the possibility of lameness) or external bleeding.
- the primary camera i.e. the visual camera
- additional cameras have been collectively referred to as cameras or "at least one camera”.
- Lighting is to be understood as any light source that can provide an increase in the quality of the image from a visual camera.
- This can be a lighting located directly in the horse's stall, e.g. a lamp connected to a visual camera, but also a lighting located outside the stall, provided that it increases horse's visibility and allows to increase the accuracy of the measurement.
- Fluorescent lamps, LED bulbs, but also infrared illuminators and other implementable light sources can be used.
- an alarm device is a device that is designed to emit an alarm signal, such as, for example, an audible signal (including a voice message), vibration or light.
- an alarm signal such as, for example, an audible signal (including a voice message), vibration or light.
- Such a device may be mounted at the location of a horse or at the location of horse’s handler. It can also be understood as a portable device of the horse’s handler, such as a mobile phone, computer or tablet, having an application containing a user interface installed on it, in which case the alarm signal is provided via components built into the horse’s handler device.
- the processing device including the number of horse’s handlers that may have the application installed in order to receive an alarm signal.
- horse’s handler may be a veterinarian, veterinary technician, horse’s owner, employee of an equine hospital, person leasing a horse, or any person who has been entrusted with the care of a horse.
- An exemplary implementation of the method of the invention in a variant with an application equipped with user interface is shown in Fig. 1-3, where Fig. 1 shows in general a model of communication with user via the user interface, Fig. 2 additionally shows the possibility of conducting a two-way communication, and Fig. 3 shows an exemplary view of the user interface.
- the present invention also relates to a system for predicting and detecting particular conditions of the horse, including, in particular, predicting the date of foaling.
- a system comprises at least one visual camera and a processing device having: • an input where an image of said horse is received from at least one visual camera,
- an artificial intelligence model such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data.
- All of the system components are configured and programmed to: a) determine the following parameters by means of image analysis:
- the system is thus adapted to perform the method described above in this document.
- Example 1 correlation of mare's activity with the upcoming moment of delivery
- a study of activity was performed for a group of mares in advanced pregnancy in order to identify differences in the activity of each mare immediately before delivery.
- the study was based solely on the analysis of the image from a visual camera using a neural network in a processing device. Randomly selected and averaged data on the change in activity of mares from the last 12 hours before delivery are presented. For mare 5, in contrast to other mares in the group, a decrease in activity before delivery was observed - a usually active mare became much less active immediately before foaling. This pattern of behaviour was atypical for the study group and was therefore dismissed when calculating the average value for the group.
- the percentages presented in the table below refer to the measurement for the same mare compared to the result from the previous day.
- Mare 5 is the exception. In this case, the mare had an initial activity which was above average. In case of very active mares, the upcoming delivery may cause a decrease in activity with the approach of the delivery moment.
- Example 2 correlation of duration of lying down of a mare with the upcoming moment of delivery
- a study of duration of lying down was performed for a group of mares in advanced pregnancy in order to identify differences in the duration of lying down of each mare immediately before delivery.
- the study was based solely on the analysis of the image from a visual camera using a neural network in a processing device. Randomly selected and averaged data on the change in duration of lying down of mares from the last 12 hours before delivery are presented. For mare 4, in contrast to other mares in the group, an increase in the duration of lying down before delivery was observed - a mare for which the duration of lying down was usually short, was lying down for longer periods of time immediately before foaling. This pattern of behaviour was atypical for the study group and was therefore dismissed when calculating the average value for the group.
- a study of duration of breaks between periods of lying down was performed for three mares in advanced pregnancy in order to identify differences in the duration of breaks between periods of lying down of each mare immediately before delivery.
- the study was based solely on the analysis of the image from a visual camera using a neural network in a processing device. Randomly selected and averaged data on the change in duration of breaks between periods of lying down of mares from the last 12 hours before delivery are presented. For mare 4, in contrast to other mares in the group, an increase in the duration of breaks between periods of lying down before delivery was observed - this trend changed immediately before foaling. This pattern of behaviour was atypical for the study group and was therefore dismissed when calculating the average value for the group.
- the region of interest in the image from the visual camera - i.e. the area of the part of abdomen which moves during breathing, is marked. This area was determined automatically using the segmentation mechanism. The effectiveness of this mechanism was investigated using two metrics, and the following results were obtained:
- the low value of the AP metric for the new set of horses is a result of the fact that the model determines many detection proposals with varying degrees of certainty, which means that it probably returned an incorrect ROI many times more than the correct one. At this stage, however, the ROI with the highest degree of certainty is always selected. In Figs. 4 and 5, the ROI with the highest certainty for a given photo can be observed (white area). mloU metric gives a better reflection of the effectiveness as it is calculated only on the already selected, most reliable ROI prediction for each photo.
- Example 5 determination of the number of breaths of a horse
- Determination of ROI was performed as described in Example 4. Inside the determined ROI, cyclic variations in light reflection were investigated, which were correlated with abdominal movement - one cyclic abdominal movement was marked as a single breath. The number of detected breaths in time was counted.
- the accuracy of the measurement is improved when the horse is not moving, ⁇ The accuracy of the measurement is improved when the horse is breathing deeply/quickly (e.g. immediately after the foal is delivered into the world), which is good news in case of horses with respiratory problems.
- segmentation mechanisms for breath count examination indicated in Examples 4 and 5 are not limited to these examples only, and may generally be used throughout the invention. Other segmentation mechanisms may also be implemented in the present invention for examining the number of breaths.
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Environmental Sciences (AREA)
- Biophysics (AREA)
- Animal Husbandry (AREA)
- Biodiversity & Conservation Biology (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention relates to a method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, in which, by means of a processing device, by means of image analysis, the following is determined: the number of breaths of the horse per minute, whether the horse is standing or lying down, a change in the level of activity of the horse. Then the obtained parameter values are analysed using an artificial intelligence model, such as a neural network, and it is determined whether the detected condition of the horse is a particular condition, wherein in case the particular condition is detected, at least one change of the output state of the processing device is made. The invention also relates to a system for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, comprising at least one visual camera and a processing device, and wherein the system is configured and programmed to determine the following by means of image analysis: number of breaths of the horse per minute, whether the horse is standing or lying down, a change in the level of activity of the horse. Then the obtained parameter values are analysed using an artificial intelligence model, such as a neural network, and it is determined whether the detected condition of the horse is a particular condition, wherein in case the particular condition is detected, at least one change of the output state of the processing device is made. The invention also relates to a system configured and programmed to carry out the above method.
Description
A method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the method
The invention relates to a method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the method.
State of the art
Methods for detecting abnormalities in the behaviour of animals from a camera image are known in the art, including those abnormalities related to the impending delivery or deterioration of the animal's health. Japanese patent application JP2020014421 A relates to a livestock birth prediction system using a image acquisition unit for acquiring image of a pre-partum animal, a monitoring unit for monitoring animal's behaviour based on an image from the image acquisition unit and a predictive unit for predicting the delivery date based on animal's behaviour. Analysis of behaviour is a dynamic analysis and is conducted based on frames obtained at specific time intervals. The obtained results are compared with the previous behaviour pattern of a given animal. Parameters such as duration of activity, standing or eating by the animal are taken into account. The system takes into account changes in these behaviours. The system is also provided with means for notifying the animal’s handler of the expected moment of delivery. The international patent application WO2021237144A1 describes a system equipped with a camera which enables determining health condition of an animal based on an image from the camera. This image is used by a neural network which compares it with models available for a given animal and other animals of the same species, and allows to determine the health condition of a given animal. Camera images are images showing depth, preferably from a near-IR camera, but the use of a conventional optical camera has also been mentioned.
A similar solution is described in the Australian patent application AU2013350867A1 , in which a camera is also used to determine health condition of a subject based on the calculations of the processor which compares the obtained images with the data available in the database and determines the health of an individual (using models, artificial intelligence and neural networks). The device can then generate information about the health condition - e.g. mastitis, fever, etc. The device is based on parameters related to temperature, but also on the observation of
given features of an individual (e.g. an animal, including a horse) or its activity - heartbeats (additional measuring device), number of breaths (additional measuring device), movements. The device also transmits alarm messages.
Document by Nabenishi H, Negishi N, Yamazaki A. entitled predicting the start of calving in Japanese Black cattle using camera image analysis” describes the usefulness of a camera for determining changes in cattle's behaviour during calving. The device is designed to predict the beginning moment of labour in cattle. Thermal analysis and image analysis alone were used. The changes in the animal's position and the duration of tail raising, as well as the movement of animals were investigated.
European patent EP3337422B1 describes prediction of calving moment based on images from a 3D camera. The image from the camera is processed and the parameters characteristic for the calving are determined, and then its time is predicted. The parameters determined include: body contractions, position (orientation) of a cow, distance between pin bones (bones within cow's pelvis), lying down of a cow, and tail movements. The obtained data is compared with the data in the database to determine the moment of calving. The system is also equipped with an alarm system to alert about upcoming delivery.
Object of the invention
The aim of the invention is to develop a solution that would enable accurate and precise, and at the same time versatile, detection and prediction of a number of particular conditions of a horse, including in particular foaling of a mare, using only images from a visual camera for measurements. At the same time, the aim of the invention was to provide a system that would allow to obtain an automatic response of external devices to the detected situation, which would contribute to increasing the safety of horses while limiting the involvement of horse’s handler.
With the present invention, the above goals were achieved.
The essence of the invention
The subject of the invention is a method for predicting and detecting particular conditions of a horse, in particular for predicting the moment of foaling. In this method a processing device having: an input where an image of said horse is received from at least one visual camera, an output which can assume at least two different output states, and an artificial intelligence model, such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data,
is used.
By means of said processing device the following steps are performed: a) by means of image analysis the following parameters are determined:
• number of breaths of the horse per minute,
• whether the horse is standing or lying down,
• a change in the level of activity of the horse, b) the obtained parameter values are analysed using the artificial intelligence model, c) as a result of this analysis, it is determined whether the detected condition of the horse is a particular condition, d) in case the particular condition is detected, at least one change of the output state of the processing device is made.
Preferably, in the method of the invention the change of the output state of the processing device causes activation of at least one accessory device selected from: additional camera, lighting, alarm device, actuator, electric lock and communication module.
In one of the variants, in step a) additionally at least one parameter selected from: the frequency of passing of manure, the shape of the stool, the frequency of urination, pawing with front leg or legs, the frequency of horse looking at its sides, the frequency of kicking the hind legs of the horse up towards the abdomen, the frequency of rolling, the frequency of eating and the frequency of drinking, is determined.
Preferably, the image from the visual camera is further analysed for the occurrence of abnormal horse body positions, such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
Preferably, the change of the output state of the processing device results in activation of an alarm device.
More preferably, the change of the output state of the processing device results in turning on a light at the location of the horse or activating at least one additional camera in the horse’s stall.
Preferably, depending on the specific detected particular condition of the horse, one or more of the following actions is initiated: a. if the horse’s foaling is predicted or detected, the surface area of the horse's stall is automatically increased or an extra portion of bedding is provided to additionally make the horse’s stall, b. if dyspnoea is predicted or detected in the horse, the stall is automatically opened to allow the horse to leave the stall, the ventilation system at the
location of the horse is activated, or drugs or supplements are automatically dosed, c. if colic, fewer or a disorder of passing of manure or urination in the horse is predicted or detected, drugs or supplements are automatically dosed.
The subject of the invention is also a system for predicting and detecting particular conditions of a horse, in particular for predicting the moment of foaling. The system comprises at least one visual camera and a processing device having:
• an input where an image of said horse is received from at least one visual camera,
• an output which can assume at least two different output states,
• and an artificial intelligence model, such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data.
The system components are configured and programmed to: a) determine the following parameters by means of image analysis:
• number of breaths of the horse per minute,
• whether the horse is standing or lying down,
• a change in the level of activity of the horse, b) analyse the obtained parameter values using the artificial intelligence model, c) as a result of this analysis, determine whether the detected condition of the horse is a particular condition, d) in case the particular condition is detected, make at least one change of the output state of the processing device.
Preferably the system further comprises one or more accessory devices selected from: additional camera, lighting, alarm device, actuator, electric lock and communication module, configured and programmed such that the change of the output state of the processing device results in activation of one or more of said accessory devices.
In one of the variants, the system is additionally configured and programmed to determine at least one parameter selected from: the frequency of passing of manure, the shape of the stool, the frequency of urination, pawing with front leg or legs, the frequency of horse looking at its sides, the frequency of kicking the hind legs of the horse up towards the abdomen, the frequency of rolling, the frequency of eating and the frequency of drinking.
Preferably the system is additionally configured and programmed to analyse the image from the visual camera for the occurrence of abnormal horse body positions,
such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
More preferably the system is additionally equipped with an alarm device, and is configured and programmed such that the change of the output state of the processing device results in activation of the alarm device.
Preferably the system is additionally configured and programmed such that the change of the output state of the processing device results in turning on a light at the location of the horse or activating at least one additional camera in the horse’s stall.
In one of the variants, the system is additionally configured and programmed such that, depending on the specific detected particular condition of the horse, one or more of the following actions is further initiated: a. if the horse’s foaling is predicted or detected, the surface area of the horse's stall is automatically increased or an extra portion of bedding is provided to additionally make the horse’s stall, b. if dyspnoea is predicted or detected in the horse, the stall is automatically opened to allow the horse to leave the stall, the ventilation system at the location of the horse is activated, or drugs or supplements are automatically dosed, c. if colic, fewer or a disorder of passing of manure or urination in the horse is predicted or detected, drugs or supplements are automatically dosed.
The subject of the invention is also the above described system, configured and programmed to carry out the above described method.
Brief description of the figures
The subject of the invention in an exemplary embodiment is illustrated in the drawing, where:
Fig. 1 shows schematically an implementation of the method of invention in one embodiment, in which the output state triggers the activation of a communication module, such as for example a Wi-Fi module.
Fig. 2 shows schematically additional functionality of the method in which the communication module may be a module providing a two-way communication between horse’s handler and the processing device in order to obtain data about the horse from the processing device or at least one camera via user interface.
Fig. 3 shows an exemplary implementation of a user interface.
Fig. 4 shows an exemplary embodiment of measuring the number of breaths of a horse according to the invention.
Fig. 5 shows another embodiment of measuring the number of breaths of a horse according to the invention.
Fig. 6 shows the results of measuring the number of breaths of a horse according to the invention.
Advantages of the invention
The method and system of the present invention for predicting and detecting particular conditions of a horse provide the possibility of a quick, automatic response of the system to the detection of a life-threatening or health-endangering condition of an animal, or a condition that could potentially have serious consequences for the horse or might require veterinary intervention. The possibility of producing various output states at the output of the system provides the possibility of activating a suitable device depending on a specific, detected condition, thus minimizing the need for often time-consuming interference of the horse’s handler. This substantially maintenance- free system increases safety of the horse, which is provided with quick, often automatic help, and at the same time increases safety of the user by minimizing the need for human to be in the presence of animal, which, especially in health- endangering or life-threatening situations, may behave in a dangerous manner.
At the same time, the system does not eliminate human being - horse's handler - from a series of taken actions. On the contrary, by means of the method and system of the invention, when a particular condition of a horse is detected, an alert informing about detection of such condition is immediately sent to the horse’s handler. This allows for a quick human intervention, if needed.
One should also not forget about the versatility of the method and system of the invention, resulting, amongst others, from the use of the functionality of analysing the number of breaths of a horse, implemented in the system and method. The method and system can be used to detect both the upcoming delivery in mares and to predict with high accuracy, even up to approximately 90%, when such delivery will take place (both short-term and long-term), as well as to detect conditions such as: problems with urination or passing of manure, disorders connected with passing of manure, colic conditions, and even respiratory disorders, especially important in horses with RAO (recurrent obstructive pulmonary disease), and even fever.
The above advantages provide that the system and method are suitable for applications that go beyond simply monitoring the overall health condition of a horse and detecting labour.
A particular advantage of the present invention is the fact that for the correct and reliable operation of the method and system of the invention, a camera providing only image data, i.e. a visual camera, is required. This feature allows to easily implement
the method of the invention in facilities already equipped with monitoring, such as industrial monitoring, without the need to install accessory equipment for collecting additional information about the horse. The method of the invention is therefore also completely non-invasive for the horse. Detailed description of the Invention
The invention relates to a method for predicting and detecting particular conditions of a horse. For the purposes of the present invention, a particular condition of a horse is defined as a condition of a horse that deviates from the normal behaviour or health condition of a particular individual. It should be realized that a given behaviour which is an indication of a particular condition for one individual may not be an indication of a particular condition for another individual, which exhibits given type of behaviour with a substantially constant frequency. An example is the lying down of horses - some of them lie down very often and/or for a long time, while others practically do not lie down at all. Exemplary particular conditions detectable or predictable by the method of the present invention will be described below in this document in details.
The detection and prediction of particular conditions of a horse are performed by means of a processing device. For the purposes of the present invention, a processing device is to be understood as any device that enables processing of an image from a visual camera in order to analyse and measure at least one of selected features. The processing device has an input (by means of which image data from at least one visual camera connected thereto is provided) and an output. Preferably, such a processing device is placed at the location of the horse, such as, for example, stall, stable, standing stall, run pen, hospital, corral (the use of one of these terms herein does not exclude implementation of the described feature with respect to other, even not directly mentioned locations if such an implementation is physically possible), and is connected by a wire or wirelessly to the at least one visual camera. A wired connection is preferred, since it minimizes the risk of delays in operation. For the purposes of the present invention, a visual camera is generally defined as a camera that collects image-only data, however, in alternative embodiments, other types of cameras, such as audio-visual cameras, may be used.
The output from the processing device is adapted to generate at least two output states and is connected to at least one external device adapted to receive the output state and to perform a particular action depending on the particular received at least one output state. For the purposes of the present invention, an output state is to be understood as a logical state such as true/false or 0/1 , a change in the position of a component of the processing device, a change in the voltage at the output of the processing device, the appearance or disappearance of a signal at the output of the
processing device - in particular such as electrical, radio or optical signal - the output state can be a specific physical value - electrical (voltage, current, impedance, capacity), radio or optical. In case more than one external device is to be used with the processing device, the output state may be communicated to the corresponding external device via an interpreter which interprets the output state or change of the output state and communicates the interpreted state to the corresponding external device. Alternatively, each of the external devices may be equipped with an individual interpreter. Such solutions, including, inter alia, communication ports would be apparent to one skilled in the art.
Examples of external devices that may be used in the present invention are: an additional camera, lighting, alarm device, actuator, electric lock and communication module. The communication module may be, for example, a GSM module, a Wi-Fi module, a Bluetooth module and other such modules. The actuator may be, for example, a power switch for air conditioning or ventilation, a power switch for a drug dispenser, and any other suitable device for carrying out a given physical action. Exemplary implementations of actuators will be described hereinbelow.
Independently and without detriment to the described system, it should be mentioned that a user terminal, such as a mobile phone, tablet or computer, can be directly connected to the processing device to monitor condition of an animal. For example, in addition to the external devices indicated above, each of the stalls may be equipped with a suitable device enabling the user to review parameters relating to the condition of an animal.
The processing device is furthermore equipped with an artificial intelligence model which is designed to analyse an image from at least one visual camera and interpret it in order to detect the particular condition of a horse. In a particularly preferred embodiment, the artificial intelligence model are neural networks, even more preferably properly trained neural networks, however, it should be understood that any artificial intelligence model may be used. Random forests are an alternative example of an artificial intelligence model. The basis of the method of the invention is to use an artificial intelligence model and to teach it the behaviour of a specific horse in order to detect abnormalities in animal’s behaviour which may be indicative of an occurrence of a particular condition. In order to detect a specific particular condition, the image from the at least one visual camera is analysed in a first step (step a)) of the method in order to measure the following parameters:
• number of breaths of a horse per minute,
• whether a horse is standing or lying down; this measurement is then used to determine the duration of lying down and the duration of breaks between periods of lying down,
• a change in the level of activity. In a preferred embodiment, a measurement of at least one parameter selected from: the frequency of passing of manure, the shape of the stool (in order to detect whether the horse has diarrhoea), the frequency of urination, pawing with front leg or legs by the horse, the frequency of horse looking at its sides, the frequency of kicking the hind legs of the horse up towards the abdomen, the frequency of rolling, the frequency of eating and the frequency of drinking, is additionally carried out.
In another preferred embodiment, the specific parameters related to the occurrence of abnormal horse body positions, such as caused by the horse's leg being stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse, may be additionally analysed.
In the next step (step b)) of the method the obtained parameter values are analysed by means of an artificial intelligence model in order to detect condition of a horse, and then in step c) it is determined whether the detected condition of the horse is a particular condition by comparing the obtained parameter values with at least one particular condition of a horse. Due to the fact that the evaluation of condition of a horse takes into account, among other things, changes in the frequency of occurrence of behaviours, it should be taken into account that the change may be caused by change of season, weather, temperature of the environment and a number of other parameters, and it should be interpreted with a certain margin of permissible fluctuations.
One of the most important particular conditions is the foaling, wherein the term "foaling" is intended to describe not only the duration of the birthing action, but also the period of preparations of the horse's body for puerperium and the period of pregnancy, during which, in the image analysis, any changes in the course of pregnancy are noticeable. It is also worth noting that the method of the invention also makes it possible to observe and analyse vital parameters of the new-born offspring during the first hours or even weeks of their life. Such measurements may contribute to the detection of early abnormalities of neonatal and adolescent age of foals, which will enable quick detection of e.g. low vital activity or overlong postpartum lying (and thus, amongst others, insufficient duration of food intake by the offspring), and, in later stages of life, of other health problems of the foal.
Other particular conditions that can be detected by the method of the present invention include, but are not limited to: problems with digestive system (colic, chewing problems, diarrhoea), problems with urinary tract / urination, respiratory problems (e.g. exacerbation of RAO symptoms, dyspnoea), fever, as well as unusual particular conditions, such as: collapsing of the horse, the horse's halter catching on the elements of stall or stall’s equipment, the horse's front or hind limbs being blocked (most often as a result of its lying in the stall in such a way that it cannot stand up on its own) or stuck in the elements of stall or stall’s equipment (e.g. hay nets or drinkers). It should be taken into account that the above list is not exhaustive and one skilled in the art will be able to easily extend it to further health problems and particular conditions which could be observed in an image from visual camera.
In order to determine the occurrence of several particular conditions, different parameters are used. The specific parameters that are used to detect and predict specific particular conditions will be presented below:
The procedure for detection of the parameters based on an image from at least one visual camera will be exemplified on the basis of analysis of number of breaths of a horse, also referred to as respiratory rate and typically given as number of breaths per minute.
The procedure for analysing the number of breaths of a horse is as follows:
• Mark the region of interest (ROI) in the image from visual camera - i.e. the area of the part of abdomen which moves during breathing (see white areas inside boxes in Fig. 4 and 5).
• Inside this region, cyclic variations in light reflection are investigated, which are correlated with abdominal movement - an abdominal movement is marked as a breath.
• The number of detected breaths in time is counted.
• Optionally, the most probable number of breaths and the certainty of the returned result are determined.
In turn, in order to determine the general activity of the horse, based on an image from the visual camera, it is determined to what extent the position of the horse's body changes in the subsequent frames of the image. The procedure for analysing activity comprises the following steps:
• A segmentation-type algorithm is run for each frame of the video material from the visual camera.
• The contour of the horse is determined.
• By determining the appropriate matrix convolution operations, the level of changes between frames is determined.
• The value is scaled to a specified activity scale, averaged over a series of frames, and returned as the activity level.
A different algorithm which analyses horse's position for each frame (also used to assess whether the horse is standing or lying down) can also be used for this purpose. The analysis may be based on determining contour of the horse and observing changes of the contour in individual frames or on determining the position of individual parts of the horse's body (e.g. limbs) and examining their movement in subsequent frames.
Based on the latter variant, specific horse movements are detected, such as turning its head (looking at its sides), rolling, etc. - the module is trained to detect specific movement patterns.
After the module is trained and taught of the behaviour patterns for a specific horse, but also for horses as a species, one can implement detection of even less typical behaviours resulting from, for example:
• Horse's halter catching on elements of a stall or stall’s equipment (e.g. stall’s rungs, drinker, manger etc.),
• Collapsing of the horse,
• Horse's leg being stuck in the elements of a stall or stall’s equipment (e.g. hay net, manger, stall’s rungs etc.),
Behaviours accompanying such particular conditions will clearly differ from normal behaviour of the horse.
The last of the mandatory parameters to be determined serves to determine whether the animal is standing or lying down at the moment. For this purpose, a binary classifier was used. The implemented algorithm detects the horse's position in a given frame of the video material and classifies it as a standing / lying down position, respectively.
In the last step of the method of the invention, as a result of detection of a particular condition, at least one change of the output state is made at the output of the processing device. The change of the output state is communicated to an external device, which activates a programmed action of the external device - each at least one change of the output state is associated with a given, specific action of at least one external device. Exemplary actions for exemplary external devices will be presented below:
For the purpose of the present invention, the additional camera is to be defined the same way as the visual camera defined hereinabove. However, it may be an audio visual camera or any other camera, preferably located at a different angle and/or height than the primary camera, which allows to view the animal from different
directions. In a particularly advantageous variant, such a camera is a thermal imaging camera, which additionally enables the estimation of the animal's body temperature (for example: temperature increase - disease, temperature drop as low as the ambient temperature - death), as well as the observation of local inflammations (e.g. inflammation of the limb, indicating the possibility of lameness) or external bleeding.
For the purposes of the present invention, the primary camera, i.e. the visual camera, and additional cameras have been collectively referred to as cameras or "at least one camera".
Lighting is to be understood as any light source that can provide an increase in the quality of the image from a visual camera. This can be a lighting located directly in the horse's stall, e.g. a lamp connected to a visual camera, but also a lighting located outside the stall, provided that it increases horse's visibility and allows to increase the accuracy of the measurement. Fluorescent lamps, LED bulbs, but also infrared illuminators and other implementable light sources can be used.
For the purpose of the present invention, an alarm device is a device that is designed to emit an alarm signal, such as, for example, an audible signal (including a voice message), vibration or light. Such a device may be mounted at the location of a horse or at the location of horse’s handler. It can also be understood as a portable device of the horse’s handler, such as a mobile phone, computer or tablet, having an application containing a user interface installed on it, in which case the alarm signal is provided via components built into the horse’s handler device. There is no limit to the number of alarm devices supported by the processing device, including the number of horse’s handlers that may have the application installed in order to receive an alarm signal.
For the purpose of the present invention, horse’s handler may be a veterinarian, veterinary technician, horse’s owner, employee of an equine hospital, person leasing a horse, or any person who has been entrusted with the care of a horse. An exemplary implementation of the method of the invention in a variant with an application equipped with user interface is shown in Fig. 1-3, where Fig. 1 shows in general a model of communication with user via the user interface, Fig. 2 additionally shows the possibility of conducting a two-way communication, and Fig. 3 shows an exemplary view of the user interface.
The present invention also relates to a system for predicting and detecting particular conditions of the horse, including, in particular, predicting the date of foaling. Such a system comprises at least one visual camera and a processing device having:
• an input where an image of said horse is received from at least one visual camera,
• an output which can assume at least two different output states,
• and an artificial intelligence model, such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data.
All of the system components are configured and programmed to: a) determine the following parameters by means of image analysis:
• number of breaths of the horse per minute,
• whether the horse is standing or lying down,
• a change in the level of activity of the horse, b) analyse the obtained parameter values using the artificial intelligence model, c) as a result of this analysis, determine whether the detected condition of the horse is a particular condition, d) in case the particular condition is detected, make at least one change of the output state of the processing device.
The system is thus adapted to perform the method described above in this document.
Examples
Example 1 - correlation of mare's activity with the upcoming moment of delivery
A study of activity was performed for a group of mares in advanced pregnancy in order to identify differences in the activity of each mare immediately before delivery. The study was based solely on the analysis of the image from a visual camera using a neural network in a processing device. Randomly selected and averaged data on the change in activity of mares from the last 12 hours before delivery are presented. For mare 5, in contrast to other mares in the group, a decrease in activity before delivery was observed - a usually active mare became much less active immediately before foaling. This pattern of behaviour was atypical for the study group and was therefore dismissed when calculating the average value for the group. The percentages presented in the table below refer to the measurement for the same mare compared to the result from the previous day.
Conclusions: In case of most of the mares in the study, there is a clear upward trend in terms of their activity - the activity increased with the approach of the moment of delivery. Mare 5 is the exception. In this case, the mare had an initial activity which was above average. In case of very active mares, the upcoming delivery may cause a decrease in activity with the approach of the delivery moment.
Example 2 - correlation of duration of lying down of a mare with the upcoming moment of delivery
A study of duration of lying down was performed for a group of mares in advanced pregnancy in order to identify differences in the duration of lying down of each mare immediately before delivery. The study was based solely on the analysis of the image from a visual camera using a neural network in a processing device. Randomly selected and averaged data on the change in duration of lying down of mares from the last 12 hours before delivery are presented. For mare 4, in contrast to other mares in the group, an increase in the duration of lying down before delivery was observed - a mare for which the duration of lying down was usually short, was lying down for longer periods of time immediately before foaling. This pattern of behaviour was atypical for the study group and was therefore dismissed when calculating the average value for the group. The percentages presented in the table below refer to the measurement for the same mare compared to the result from the previous day.
Conclusions: In case of most of the mares in the study, there is a clear downward trend in terms of duration of lying down - the duration of lying down decreased with
the approach of the moment of delivery. Mare 4 is the exception. In this case, the mare had an initial duration of lying down which was below average. In case of mares for which the duration of lying down is short, the upcoming delivery may cause an increase in duration of lying down with the approach of the delivery moment. Example 3 - correlation of duration of breaks between periods of lying down of a mare with the upcoming moment of delivery
A study of duration of breaks between periods of lying down was performed for three mares in advanced pregnancy in order to identify differences in the duration of breaks between periods of lying down of each mare immediately before delivery. The study was based solely on the analysis of the image from a visual camera using a neural network in a processing device. Randomly selected and averaged data on the change in duration of breaks between periods of lying down of mares from the last 12 hours before delivery are presented. For mare 4, in contrast to other mares in the group, an increase in the duration of breaks between periods of lying down before delivery was observed - this trend changed immediately before foaling. This pattern of behaviour was atypical for the study group and was therefore dismissed when calculating the average value for the group. The percentages presented in the table below refer to the measurement for the same mare compared to the result from the previous day.
Conclusions: In case of most of the mares in the study, there is a clear downward trend in terms of duration of breaks between periods of lying down - the duration of breaks between periods of lying down decreased with the approach of the moment of delivery. Mare 4 is the exception. In this case, the observed initial duration of breaks between periods of lying down for the mare was short (below average) and the duration of those breaks was increasing with the approach of the delivery moment. In case of mares for which the duration of breaks between periods of lying down is short, the upcoming delivery may cause an increase in duration of those breaks between periods of lying down with the approach of the delivery moment.
Example 4 - determination of a suitable segmentation mechanism metric for the study of number of breaths of a horse
The region of interest in the image from the visual camera - i.e. the area of the part of abdomen which moves during breathing, is marked. This area was determined automatically using the segmentation mechanism. The effectiveness of this mechanism was investigated using two metrics, and the following results were obtained:
• for horses during the training phase of the artificial intelligence model: o AP - 87.13 o mloU - 96.94
• for new horses not seen before by the artificial intelligence model: o AP - 20.2 o mloU - 86.33.
These results were considered accurate enough to use the automatic selection of the region of interest.
Conclusions: The low value of the AP metric for the new set of horses is a result of the fact that the model determines many detection proposals with varying degrees of certainty, which means that it probably returned an incorrect ROI many times more than the correct one. At this stage, however, the ROI with the highest degree of certainty is always selected. In Figs. 4 and 5, the ROI with the highest certainty for a given photo can be observed (white area). mloU metric gives a better reflection of the effectiveness as it is calculated only on the already selected, most reliable ROI prediction for each photo.
Example 5 - determination of the number of breaths of a horse
Determination of ROI was performed as described in Example 4. Inside the determined ROI, cyclic variations in light reflection were investigated, which were correlated with abdominal movement - one cyclic abdominal movement was marked as a single breath. The number of detected breaths in time was counted.
The study was conducted under the following conditions:
• the animal had no restrictions regarding its movement - it was moving naturally,
• a visual camera was used,
• no approximations of the obtained values were used,
• the region moving during breathing was marked automatically (no human interference),
• the breaths were visible to the human observing the image from the visual camera (important for obtaining the control value).
The average values of breath count results for a group of horses are shown in Fig. 6 and summarized below:
• Actual number of breaths: 32-33 breaths per minute
• The number of breaths obtained with the model: 32 breaths per minute, · Accuracy of the obtained result: 95% on average.
General conclusions from the study:
• The accuracy of the measurement decreases with the decrease in the resolution of the visual camera,
• The accuracy of the measurement is improved when the horse is not moving, · The accuracy of the measurement is improved when the horse is breathing deeply/quickly (e.g. immediately after the foal is delivered into the world), which is good news in case of horses with respiratory problems.
The segmentation mechanisms for breath count examination indicated in Examples 4 and 5 are not limited to these examples only, and may generally be used throughout the invention. Other segmentation mechanisms may also be implemented in the present invention for examining the number of breaths.
Claims
1. A method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, in which, by means of a processing device having:
• an input where an image of said horse is received from at least one visual camera,
• an output which can assume at least two different output states,
• and an artificial intelligence model, such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data, in which the following steps are performed: a) by means of image analysis the following parameters are determined:
• number of breaths of the horse per minute,
• whether the horse is standing or lying down,
• a change in the level of activity of the horse, b) the obtained parameter values are analysed using the artificial intelligence model, c) as a result of this analysis, it is determined whether the detected condition of the horse is a particular condition, d) in case the particular condition is detected, at least one change of the output state of the processing device is made.
2. The method according to claim 1 , wherein the change of the output state of the processing device causes the activation of at least one accessory device selected from: additional camera, lighting, alarm device, actuator, electric lock and communication module.
3. The method according to claim 1 or 2, in which in step a) additionally at least one parameter selected from: the frequency of passing of manure, the shape of the stool, the frequency of urination, pawing with front leg or legs, the frequency of horse looking at its sides, the frequency of kicking the hind legs of the horse up towards the abdomen, the frequency of rolling, the frequency of eating and the frequency of drinking, is determined.
4. The method according to any one of claims 1 to 3, wherein the image from the visual camera is further analysed for the occurrence of abnormal horse body positions, such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
5. The method according to any one of claims 1 to 4, wherein the change of the output state of the processing device results in activation of an alarm device.
6. The method according to any one of claims 1 to 5, wherein the change of the output state of the processing device results in turning on a light at the location of the horse or activating at least one additional camera in the horse’s stall.
7. The method according to any one of claims 1 to 6, wherein depending on the specific detected particular condition of the horse one or more of the following actions is initiated: a. if the horse’s foaling is predicted or detected, the surface area of the horse's stall is automatically increased or an extra portion of bedding is provided to additionally make the horse’s stall, b. if dyspnoea is predicted or detected in the horse, the stall is automatically opened to allow the horse to leave the stall, the ventilation system at the location of the horse is activated, or drugs or supplements are automatically dosed, c. if colic, fewer or a disorder of passing of manure or urination in the horse is predicted or detected, drugs or supplements are automatically dosed.
8. A system for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, comprising at least one visual camera and a processing device having:
• an input where an image of said horse is received from at least one visual camera,
• an output which can assume at least two different output states,
• and an artificial intelligence model, such as a neural network, coupled to the input and receiving parameters derived from the image of said horse as input data, and coupled to the output, and trained to predict and detect particular conditions of the horse from such input data, whereby the system components are configured and programmed to: a) determine the following parameters by means of image analysis:
• number of breaths of the horse per minute,
• whether the horse is standing or lying down,
• a change in the level of activity of the horse, b) analyse the obtained parameter values using the artificial intelligence model, c) as a result of this analysis, determine whether the detected condition of the horse is a particular condition,
d) in case the particular condition is detected, make at least one change of the output state of the processing device.
9. The system according to claim 8 further comprising one or more accessory devices selected from: additional camera, lighting, alarm device, actuator, electric lock and communication module, configured and programmed such that the change of the output state of the processing device results in activation of one or more of said accessory devices.
10. The system according to claim 8 or 9, which is additionally configured and programmed to determine at least one parameter selected from: the frequency of passing of manure, the shape of the stool, the frequency of urination, pawing with front leg or legs, the frequency of horse looking at its sides, the frequency of kicking the hind legs of the horse up towards the abdomen, the frequency of rolling, the frequency of eating and the frequency of drinking.
11. The system according to any one of claims 8 to 10, which is additionally configured and programmed to analyse the image from the visual camera for the occurrence of abnormal horse body positions, such as caused by the horse's front or hind limbs being blocked or stuck in the elements of stall or stall’s equipment, the horse's halter catching on the elements of stall or stall’s equipment, and the collapsing of the horse.
12. The system according to any one of claims 8 to 11 , which is additionally equipped with an alarm device, and which is configured and programmed such that the change of the output state of the processing device results in activation of the alarm device.
13. The system according to any one of claims 8 to 12, which is additionally configured and programmed such that the change of the output state of the processing device results in turning on a light at the location of the horse or activating at least one additional camera in the horse’s stall.
14. The system according to any one of claims 8 to 13, which is additionally configured and programmed such that depending on the specific detected particular condition of the horse one or more of the following actions is further initiated: a. if the horse’s foaling is predicted or detected, the surface area of the horse's stall is automatically increased or an extra portion of bedding is provided to additionally make the horse’s stall, b. if dyspnoea is predicted or detected in the horse, the stall is automatically opened to allow the horse to leave the stall, the ventilation system at the
location of the horse is activated, or drugs or supplements are automatically dosed, c. if colic, fewer or a disorder of passing of manure or urination in the horse is predicted or detected, drugs or supplements are automatically dosed.
15. The system according to any one of claims 8 to 14, configured and programmed to carry out the method according to any one of claims 1 to 7.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP21461550 | 2021-06-09 | ||
PCT/IB2022/055347 WO2022259183A1 (en) | 2021-06-09 | 2022-06-08 | A method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the method |
Publications (1)
Publication Number | Publication Date |
---|---|
EP4366518A1 true EP4366518A1 (en) | 2024-05-15 |
Family
ID=83004617
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP22757642.8A Pending EP4366518A1 (en) | 2021-06-09 | 2022-06-08 | A method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the method |
Country Status (2)
Country | Link |
---|---|
EP (1) | EP4366518A1 (en) |
WO (1) | WO2022259183A1 (en) |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3468353B1 (en) * | 2016-06-13 | 2023-12-27 | Equimetrics Limited | A garment |
WO2019160645A1 (en) * | 2018-02-19 | 2019-08-22 | Equus Global Holdings Llc | Mobile animal surveillance and distress monitoring |
US10905105B2 (en) * | 2018-06-19 | 2021-02-02 | Farm Jenny LLC | Farm asset tracking, monitoring, and alerts |
CA3123097A1 (en) * | 2018-10-26 | 2020-04-30 | Swinetech, Inc. | Livestock stillbirthing alerting system |
KR102139048B1 (en) * | 2020-01-22 | 2020-07-30 | 주식회사 피플멀티 | Health care monitoring system of pregnancy animal using smart health care apparatus |
-
2022
- 2022-06-08 EP EP22757642.8A patent/EP4366518A1/en active Pending
- 2022-06-08 WO PCT/IB2022/055347 patent/WO2022259183A1/en active Application Filing
Also Published As
Publication number | Publication date |
---|---|
WO2022259183A1 (en) | 2022-12-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Chapa et al. | Accelerometer systems as tools for health and welfare assessment in cattle and pigs–a review | |
Costa et al. | Symposium review: Precision technologies for dairy calves and management applications | |
Frost et al. | A review of livestock monitoring and the need for integrated systems | |
ES2951065T3 (en) | Surveillance and supervision of animal discomfort by mobile phone | |
US9538729B2 (en) | Cattle monitoring for illness | |
AU2011218640B2 (en) | Detection apparatus | |
US20200214266A1 (en) | Domestic animal information management system, domestic animal barn, domestic animal information management program, and domestic animal information management method | |
KR102035672B1 (en) | Apparatus for monitoring location and behavior of cattle and method thereof | |
EP1520471A2 (en) | A device for monitoring an animal | |
US11856924B2 (en) | Garment | |
JP2017060407A (en) | System and method for health care of ruminant | |
MX2008013342A (en) | Method and system for monitoring the condition of livestock. | |
KR101976519B1 (en) | Apparatus for monitoring ruminant stomach of cattle and method thereof | |
JPWO2016171077A1 (en) | Information processing system | |
WO2013122468A1 (en) | Automated monitoring and controlling of undesired livestock behaviour | |
KR102141582B1 (en) | Prediction method and the apparatus for onset time of sow farrowing by image analysis | |
EP3756458A1 (en) | Weight determination of an animal based on 3d imaging | |
KR102418883B1 (en) | Monitoring methods of ruminant stomach of cattle | |
Grodkowski et al. | Comparison of different applications of automatic herd control systems on dairy farms–a review | |
WO2013020825A1 (en) | Device having a sensor for measuring a vital value in an animal | |
EP4366518A1 (en) | A method for predicting and detecting particular conditions of a horse, in particular for predicting the date of foaling, and a system for carrying out the method | |
Miedema | Investigating the use of behavioural, accelerometer and heart rate measurements to predict calving in dairy cows | |
Lahart et al. | Exploring the potential of ingestive behaviour, body measurements, thermal imaging, heart rate and blood pressure to predict dry matter intake in grazing dairy cows | |
JP7410607B1 (en) | Feeding management system and feeding management method | |
Ruhil et al. | Data sensing and computation technologies for amimal welfare |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: UNKNOWN |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE |
|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE |
|
17P | Request for examination filed |
Effective date: 20240104 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR |