CN111860203A - Abnormal pig identification device, system and method based on image and audio mixing - Google Patents
Abnormal pig identification device, system and method based on image and audio mixing Download PDFInfo
- Publication number
- CN111860203A CN111860203A CN202010600412.4A CN202010600412A CN111860203A CN 111860203 A CN111860203 A CN 111860203A CN 202010600412 A CN202010600412 A CN 202010600412A CN 111860203 A CN111860203 A CN 111860203A
- Authority
- CN
- China
- Prior art keywords
- group
- pig
- image
- sample
- sound data
- 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.)
- Granted
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 102
- 238000000034 method Methods 0.000 title claims abstract description 30
- 241000282887 Suidae Species 0.000 claims abstract description 87
- 230000008859 change Effects 0.000 claims abstract description 55
- 238000007499 fusion processing Methods 0.000 claims abstract description 39
- 238000005070 sampling Methods 0.000 claims description 50
- 238000003709 image segmentation Methods 0.000 claims description 31
- 238000000605 extraction Methods 0.000 claims description 29
- 238000000926 separation method Methods 0.000 claims description 24
- 238000010801 machine learning Methods 0.000 claims description 20
- 230000008569 process Effects 0.000 claims description 16
- 230000004807 localization Effects 0.000 claims description 13
- 238000009395 breeding Methods 0.000 claims description 12
- 230000001488 breeding effect Effects 0.000 claims description 12
- 206010011224 Cough Diseases 0.000 claims description 8
- 108010001267 Protein Subunits Proteins 0.000 claims description 4
- 230000036760 body temperature Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 22
- 230000002265 prevention Effects 0.000 abstract description 8
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 239000000284 extract Substances 0.000 description 3
- 208000035473 Communicable disease Diseases 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 208000031636 Body Temperature Changes Diseases 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L17/00—Speaker identification or verification techniques
- G10L17/26—Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/80—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for detecting, monitoring or modelling epidemics or pandemics, e.g. flu
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Multimedia (AREA)
- Public Health (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Primary Health Care (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides an abnormal pig identification device, system and method based on image and audio mixing, which are used for acquiring a group image set and a group voice data of pigs in a group fence and identifying whether each pig is an abnormal pig or not according to the group image set and the group voice data through an identification model in an identification module. The identification model comprises a fusion processing sub-model, and the fusion processing sub-model outputs the identification result of whether any pig is an abnormal pig or not according to the sign change information and the individual voice data of any pig determined by the group image set and the group voice data. Through the combination of the image and the sound, the accuracy of identifying the abnormal pig is improved, and the reliability of the detection result is high. Meanwhile, manual detection is not needed, the detection efficiency of abnormal pigs is improved, epidemic prevention and control are facilitated, and loss is reduced.
Description
Technical Field
The invention relates to the technical field of machine learning, in particular to an abnormal pig identification device, system and method based on image and audio mixing.
Background
The growth and farrowing of the pigs are seriously affected by the health of the pigs, so that economic loss is brought. Early discovery and early treatment are important means for epidemic prevention and control. In the existing method, each pig is usually detected manually at regular intervals or is screened from the group raising fence through manual observation to obtain the pigs with abnormal performances. The pigs with abnormal manifestations are further detected to prevent and control possible infectious diseases through detection or treat problems existing in the pigs as soon as possible, so that loss is reduced.
Therefore, the existing method carries out epidemic situation prevention and control on the pigs through manual detection so as to reduce loss, and has low manual detection efficiency and low reliability of detection results.
Disclosure of Invention
The embodiment of the invention provides an abnormal pig identification device, system and method based on image and audio mixing, which are used for solving the problems that the existing method carries out epidemic situation prevention and control on pigs through manual detection so as to reduce loss, the manual detection efficiency is low, and the reliability of detection results is low.
In view of the above technical problems, in a first aspect, an embodiment of the present invention provides an abnormal pig identification device based on image and audio mixing, including an obtaining module and an identification module;
The acquisition module acquires a group image set obtained by carrying out infrared image acquisition and/or camera shooting on a group raising fence in which pigs are confined within a preset time period, and group sound data acquired by the pigs in the group raising fence within the preset time period;
the identification module inputs the group image set and the group sound data into an identification model to obtain an identification result of whether each pig of the group raising fence output by the identification model is an abnormal pig or not;
the identification model comprises a fusion processing submodel, and the fusion processing submodel outputs the result of whether any pig is an abnormal pig or not according to the combination of the sign change information and the individual sound data of any pig; the sign change information of any pig is determined according to the individual image set corresponding to any pig in the group image set; the individual sound data of any pig is determined according to the group sound data and the group image set.
Alternatively,
the recognition model further comprises an image segmentation sub-model, a physical sign extraction sub-model, an input sub-unit, a sound positioning sub-model and a sound sub-ion unit:
The image segmentation sub-model performs image segmentation according to the group images to obtain an individual image set corresponding to each pig; for any pig, the sign extraction sub-model is used for extracting sign change information of any pig according to the individual image set corresponding to any pig;
the input subunit is used for acquiring group sound data acquired at any time point according to the group sound data and inputting the group sound data acquired at any time point into the sound positioning sub-model;
the sound positioning sub-model determines a main sound source position corresponding to the group sound data collected at any time point according to the position of pickup equipment collecting the group sound data in the group fence and the group sound data collected at any time point; the sound separation subunit outputs individual sound data corresponding to each pig according to the main sound source position corresponding to the group sound data collected at each time point output by the sound positioning submodel and the pig position of each pig at different time points determined according to the group image set;
the fusion processing submodel outputs a result whether any pig is an abnormal pig or not according to the combination of the sign change information output by the sign extraction submodel for any pig and the individual sound data corresponding to any pig output by the sound separation submodel;
And the position of the main sound source represents the position of the pig which emits the maximum decibel sound in the group-raising fence at any time point.
Alternatively,
the sound separation subunit is to:
acquiring a single-frame image obtained by acquiring and/or shooting an infrared image and/or a camera of the group fence at any time point according to the group image set, and acquiring a main sound source position corresponding to group sound data acquired at any time point according to the sound positioning sub-model;
judging whether pigs with the distance to the main sound source position smaller than or equal to a preset distance threshold exist according to the positions of the pigs in the single-frame image, if so, taking the pigs closest to the main sound source position as positioning pigs, and taking the group sound data acquired at any time point as the group sound data corresponding to the positioning pigs;
and acquiring group sound data corresponding to any pig from the group sound data acquired at each time point in the preset time period, and denoising the group sound data corresponding to any pig to obtain individual sound data corresponding to any pig.
Alternatively,
the establishment process of the image segmentation sub-model comprises the following steps:
acquiring a plurality of group image sets obtained by carrying out infrared image acquisition and/or camera shooting on pigs in a group raising fence in a sampling time period, taking the group image sets as sample group image sets, and segmenting a sample individual image set corresponding to each pig from the sample group image sets;
and taking the sample group image set as an input, taking the sample individual image set corresponding to each divided pig as an output, and performing model training through machine learning to obtain the image division sub-model.
Alternatively,
the establishment process of the sign extraction submodel comprises the following steps:
obtaining a sample individual image set of any sample pig, wherein the sample individual image set comprises at least one of the following images: a first image subset continuously acquired during a first time period when any sample pig is in a cough state, a second image subset continuously acquired during a second time period when any sample pig is in a doze state, and an infrared image subset continuously acquired during a sampling time period;
taking the sample individual image set as input, taking sample sign change information as output, and performing model training through machine learning to obtain the sign extraction sub-model; wherein, the sample sign change information comprises at least one of the following information: information corresponding to the first subset of images that a cough has occurred in the pig sample during the first time period, information corresponding to the second subset of images that the pig sample during the second time period is dozing, and information corresponding to a change in body temperature of the pig in the infrared subset of images;
Wherein the second time period within the first time period is a sub-time period in a sampling time period for acquiring the sample group image set; the first image subset, the second image subset, and the infrared image subset are subsets of the sample individual image set.
Optionally, the establishing process of the sound localization submodel includes:
acquiring the position of each sound pickup device for acquiring group sound data of pigs in the group raising fence, and the group sound data acquired by each sound pickup device at a set sampling time point as sample group sound data; at the set sampling time point, the sample pig which sends the maximum decibel sound in the group-raising fence is only positioned at the position of the main sound source of the sample;
and taking the position of each sound pickup device for collecting the sample group sound data in the group culture bar, the sample group sound data collected at the set sampling time point as input, and taking the position of the sample main sound source as output, and performing model training through machine learning to obtain the sound positioning sub-model.
Optionally, the sound separation subunit is configured to:
acquiring a single-frame sample image obtained by performing infrared image acquisition and/or shooting on the group culture bar at the set sampling time point according to the sample group image set, and acquiring a main sound source position of a sample corresponding to group sound data acquired at the set sampling time point according to the sound positioning sub-model;
Judging whether pigs with the distance to the main sound source position of the sample being smaller than or equal to a preset distance threshold exist according to the positions of the pigs in the single-frame sample image, if so, taking the pigs closest to the main sound source position of the sample as positioning pigs, and taking the group sound data acquired at the set sampling time point as the group sound data corresponding to the positioning pigs;
and acquiring group sound data corresponding to the pig of any sample from the group sound data acquired at each sampling time point in a sampling time period, and denoising the group sound data corresponding to the pig of any sample to obtain sample individual sound data corresponding to the pig of any sample.
Optionally, the process of establishing the fusion processing sub-model includes:
acquiring an image segmentation sub-model, a physical sign extraction sub-model and a sound positioning sub-model which are obtained by performing model training through machine learning, and an input sub-unit and a sound separation sub-unit;
inputting the sample group image set into the image segmentation sub-model, and acquiring sample sign change information extracted by the sign extraction sub-model for any sample pig according to the individual image set of any sample pig in the image segmentation sub-model; the sample group image set is obtained by carrying out infrared image acquisition and/or camera shooting on the pigs in the group raising fence;
Inputting sample group sound data into the input subunit, inputting the sample group sound data collected at the set sampling time point into the sound localization sub-model by the input subunit, outputting sample individual sound data of any sample pig by the sound separation subunit according to a sample main sound source position corresponding to the group sound data collected at each sampling time point in the sampling time period output by the sound localization sub-model and a position of each pig at each sampling time point in a sample group image;
and taking the combination of the sample sign change information of any sample pig and the sample individual sound data as input, taking the result of whether any sample pig is an abnormal pig or not as output, and performing model training through machine learning to obtain the fusion processing sub-model.
In a second aspect, an embodiment of the present invention provides an abnormal pig recognition system based on image and audio mixing, including infrared cameras and/or video cameras disposed at different positions of a group raising fence, sound pickup equipment disposed at different positions of the group raising fence, and any one of the above abnormal pig recognition apparatuses based on image and audio mixing;
The infrared camera is used for carrying out infrared image acquisition on the group breeding fence with the pigs in captivity to obtain a group image set comprising infrared images, and the obtained group image set is sent to the abnormal pig identification device based on image and audio mixing;
the camera is used for shooting the group breeding fence to obtain a group image set comprising shot images, and sending the obtained group image set to the abnormal pig identification device based on image and audio mixing;
the sound pickup equipment is used for collecting sound of the group raising fence to obtain group sound data, and sending the obtained group sound data to the abnormal pig identification device based on image and audio mixing.
In a third aspect, an embodiment of the present invention provides an abnormal pig identification method based on image and audio mixing, including:
the acquisition module acquires a group image set obtained by carrying out infrared image acquisition and/or camera shooting on a group raising fence in which pigs are confined within a preset time period, and group sound data acquired by the pigs in the group raising fence within the preset time period;
the identification module inputs the group image set and the group sound data into an identification model to obtain an identification result of whether each pig of the group raising fence output by the identification model is an abnormal pig or not;
The identification model comprises a fusion processing submodel, and the fusion processing submodel outputs the result of whether any pig is an abnormal pig or not according to the combination of the sign change information and the individual sound data of any pig; the sign change information of any pig is determined according to the individual image set corresponding to any pig in the group image set; the individual sound data of any pig is determined according to the group sound data and the group image set.
The embodiment of the invention provides an abnormal pig identification device, system and method based on image and audio mixing, which are used for acquiring a group image set and a group voice data of pigs in a group fence and identifying whether each pig is an abnormal pig or not according to the group image set and the group voice data through an identification model in an identification module. The identification model comprises a fusion processing sub-model, and the fusion processing sub-model outputs the identification result of whether any pig is an abnormal pig or not according to the sign change information and the individual voice data of any pig determined by the group image set and the group voice data. Through the combination of the image and the sound, the accuracy of identifying the abnormal pig is improved, and the reliability of the detection result is high. Meanwhile, manual detection is not needed, the detection efficiency of abnormal pigs is improved, epidemic prevention and control are facilitated, and loss is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a block diagram of an abnormal pig identification device based on image and audio mixing according to an embodiment of the present invention;
fig. 2 is a schematic diagram of the principle of identifying abnormal pigs by the identification model according to another embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a block diagram of a structure of an abnormal pig recognition device based on image and audio mixing according to the present embodiment, and referring to fig. 1, the abnormal pig recognition device based on image and audio mixing includes an acquisition module 101 and a recognition module 102;
the acquisition module 101 acquires a group image set obtained by performing infrared image acquisition and/or camera shooting on a group raising fence in which pigs are housed within a preset time period, and group sound data acquired by the pigs within the group raising fence within the preset time period;
the identification module 102 inputs the group image set and the group voice data into an identification model to obtain an identification result of whether each pig in the group raising column output by the identification model is an abnormal pig;
the identification model comprises a fusion processing submodel, and the fusion processing submodel outputs the result of whether any pig is an abnormal pig or not according to the combination of the sign change information and the individual sound data of any pig; the sign change information of any pig is determined according to the individual image set corresponding to any pig in the group image set; the individual sound data of any pig is determined according to the group sound data and the group image set.
The group image set is an image set obtained by continuously acquiring images of the group fence in the preset time period, and may be an image set composed of infrared images, an image set composed of photographic images, or an image set composed of both infrared images and photographic images.
And the group sound data is obtained by continuously collecting sound of the group culture bar in the preset time period.
After the group image set and the group voice data are input into the identification model by the identification module, the identification model extracts the sign change information and the individual voice data corresponding to each pig from the group image set and the group voice data, so that the fusion processing sub-model in the identification model identifies whether each pig is an abnormal pig or not according to the combination of the sign change information and the individual voice data corresponding to each pig, provides a reference for breeding personnel whether the pig is an abnormal pig or not, is favorable for the breeding personnel to check the abnormal pig early, effectively prevents the spread of infectious diseases, and avoids economic loss.
The embodiment provides an abnormal pig identification device based on image and audio mixing, which is used for acquiring a group image set and a group voice data of pigs in a group raising fence, and identifying whether each pig is an abnormal pig or not according to the group image set and the group voice data through an identification model in an identification module. The identification model comprises a fusion processing sub-model, and the fusion processing sub-model outputs the identification result of whether any pig is an abnormal pig or not according to the sign change information and the individual voice data of any pig determined by the group image set and the group voice data. Through the combination of the image and the sound, the accuracy of identifying the abnormal pig is improved, and the reliability of the detection result is high. Meanwhile, manual detection is not needed, the detection efficiency of abnormal pigs is improved, epidemic prevention and control are facilitated, and loss is reduced.
Further, on the basis of the above embodiment, the recognition model further includes an image segmentation sub-model, a physical sign extraction sub-model, an input sub-unit, a sound localization sub-model, and a sound ionization unit;
the image segmentation sub-model performs image segmentation according to the group images to obtain an individual image set corresponding to each pig; for any pig, the sign extraction sub-model is used for extracting sign change information of any pig according to the individual image set corresponding to any pig;
the input subunit is used for acquiring group sound data acquired at any time point according to the group sound data and inputting the group sound data acquired at any time point into the sound positioning sub-model;
the sound positioning sub-model determines a main sound source position corresponding to the group sound data collected at any time point according to the position of pickup equipment collecting the group sound data in the group fence and the group sound data collected at any time point; the sound separation subunit outputs individual sound data corresponding to each pig according to the main sound source position corresponding to the group sound data collected at each time point output by the sound positioning submodel and the pig position of each pig at different time points determined according to the group image set;
The fusion processing submodel outputs a result whether any pig is an abnormal pig or not according to the combination of the sign change information output by the sign extraction submodel for any pig and the individual sound data corresponding to any pig output by the sound separation submodel;
and the position of the main sound source represents the position of the pig which emits the maximum decibel sound in the group-raising fence at any time point.
Fig. 2 is a schematic diagram of the principle of the identification model provided in this embodiment for identifying abnormal pigs, and referring to fig. 2, the image segmentation sub-model and the physical sign extraction sub-model extract physical sign change information of any pig according to the input population image set. And the input subunit, the sound positioning subunit and the sound ion separating unit extract the individual sound data of any pig according to the group image set and the group sound data. And the fusion processing sub-model realizes the identification of whether any pig is an abnormal pig or not according to the combination of the sign change information and the individual voice data of any pig.
It should be noted that the image segmentation sub-model, the physical sign extraction sub-model, and the sound localization sub-model may be models obtained by training through machine learning, or models having corresponding functions implemented through a combination of some functions, which is not specifically limited in this embodiment.
In the embodiment, the identification of the abnormal pig is realized by combining the image and the sound through the identification model consisting of the image segmentation sub-model, the physical sign extraction sub-model, the input sub-unit, the sound positioning sub-model, the sound separation sub-unit and the fusion processing sub-module, and the accuracy of identifying the abnormal pig is improved.
Further, on the basis of the above embodiments, the sound separation subunit is configured to:
acquiring a single-frame image obtained by acquiring and/or shooting an infrared image and/or a camera of the group fence at any time point according to the group image set, and acquiring a main sound source position corresponding to group sound data acquired at any time point according to the sound positioning sub-model;
judging whether pigs with the distance to the main sound source position smaller than or equal to a preset distance threshold exist according to the positions of the pigs in the single-frame image, if so, taking the pigs closest to the main sound source position as positioning pigs, and taking the group sound data acquired at any time point as the group sound data corresponding to the positioning pigs;
and acquiring group sound data corresponding to any pig from the group sound data acquired at each time point in the preset time period, and denoising the group sound data corresponding to any pig to obtain individual sound data corresponding to any pig.
Briefly, the sound separation subunit is configured to combine the position of each pig in the group image set and the main sound source position corresponding to the group sound data acquired at each time point within the preset time period determined by the sound localization sub-model, and to associate the group sound data acquired at each time point with the pig, so as to obtain the individual sound data corresponding to each pig.
The position of each pig in the single frame image can be determined according to a mark position set in the group-raising bar in advance, for example, two or more mark positions are marked in the group-raising bar in advance, and if the relative distance between each mark position in the image is known, the position of the pig in the actual group-raising bar can be determined according to the relative distance between each mark position in the image and the relative distance between the pig and any mark position in the image.
Wherein the preset distance threshold is a set value, for example, 3 meters.
If the pigs with the distance to the main sound source position smaller than or equal to the preset distance threshold do not exist, processing is not carried out, and whether the corresponding positioning pigs exist in the group sound data collected at the next time point is continuously judged until the operation of judging whether the pigs with the distance to the main sound source position smaller than or equal to the preset distance threshold exist in the group sound data collected at each time point of the preset time period is carried out.
After the position of each pig in the single-frame image at each time point and the group sound data acquired at each time point are combined, the pig corresponding to the group sound data acquired at each time point is determined, the group sound data acquired at each time point corresponding to any pig is acquired for any pig, and the reference sound data is supplemented to the time point without the group sound data corresponding to any pig, so that the individual sound data of any pig is obtained. And the individual sound data of any pig comprises reference sound data or group sound data of any pig corresponding to each time point in the preset time period. The reference sound data is the sound data collected by the normal pigs or a set sound data, the reference sound data does not affect the data characteristics expressed by the group sound data in the individual sound data, namely, the judgment process of the abnormal pigs is not affected, and the reference sound data is only used for completely supplementing the sound data in the preset time period.
In this embodiment, the sound separation subunit combines the positions of the pigs in the single-frame image and the main sound source position for positioning the group sound data, so as to separate the sound data, determine the individual sound data corresponding to each pig, and identify abnormal pigs according to the individual sound data and the sign change information.
Further, on the basis of the foregoing embodiments, the establishing process of the image segmentation sub-model includes:
acquiring a plurality of group image sets obtained by carrying out infrared image acquisition and/or camera shooting on pigs in a group raising fence in a preset sampling time period, taking the group image sets as sample group image sets, and segmenting a sample individual image set corresponding to each pig from the sample group image sets;
and taking the sample group image set as an input, taking the sample individual image set corresponding to each divided pig as an output, and performing model training through machine learning to obtain the image division sub-model.
Wherein the image segmentation sub-model may be a convolutional neural network type model.
And obtaining the image segmentation sub-model through a training process of a plurality of models by taking the sample group image sets as input and the segmented sample individual image sets corresponding to each pig as output.
In the embodiment, the training of the image segmentation sub-model is realized through machine learning, so that the image segmentation sub-model can separate an individual image set corresponding to each pig from a group image set.
Further, on the basis of the above embodiments, the establishing process of the sign extraction submodel includes:
Obtaining a sample individual image set of any sample pig, wherein the sample individual image set comprises at least one of the following images: a first image subset continuously acquired within a first time period when any sample pig is in a cough state, a second image subset continuously acquired within a second time period when any sample pig is in a doze state, and an infrared image subset continuously acquired within a preset sampling time period;
taking the sample individual image set as input, taking sample sign change information as output, and performing model training through machine learning to obtain the sign extraction sub-model; wherein, the sample sign change information comprises at least one of the following information: information corresponding to the first subset of images that a cough has occurred in the pig sample during the first time period, information corresponding to the second subset of images that the pig sample during the second time period is dozing, and information corresponding to a change in body temperature of the pig in the infrared subset of images;
wherein the second time period within the first time period is a sub-time period in a sampling time period for acquiring the sample group image set; the first image subset, the second image subset, and the infrared image subset are subsets of the sample individual image set.
Wherein the image segmentation submodel may also be a convolutional neural network type model.
The sample sign change information may be marked in advance according to the sample individual image set, for example, the sample sign change information is a sampling time period before a first time period, the pig with any sample has no abnormal sign, the pig with any sample has cough only in the first time period, and the pig with any sample has no abnormal sign in the sampling time period after the first time period.
Generally, in the sample physical sign change information, the body temperature change information extracted through the infrared image subset and the cough and/or doze occurrence information extracted through the camera image (including the first image subset and/or the second image subset) are information of two dimensions. The information of the two dimensions can be jointly used as a part of the sign change information of the sample to participate in the subsequent identification process of abnormal pigs.
Training the model through a plurality of sample individual image sets and sample sign change information extracted from the sample individual image sets to obtain a sign extraction submodel.
In the embodiment, training of the sign extraction submodel is realized through the sample sign change information extracted from the sample individual image set in advance, and the sign change information of each pig is extracted through the sign extraction submodel, so that the abnormal pig is identified based on the combination of the sign change information and the individual sound data.
Further, on the basis of the foregoing embodiments, the establishing process of the sound localization submodel includes:
acquiring the position of each sound pickup device for acquiring group sound data of pigs in the group raising fence, and the group sound data acquired by each sound pickup device at a set sampling time point as sample group sound data; at the set sampling time point, the sample pig which sends the maximum decibel sound in the group-raising fence is only positioned at the position of the main sound source of the sample;
and taking the position of each sound pickup device for collecting the sample group sound data in the group culture bar, the sample group sound data collected at the set sampling time point as input, and taking the position of the sample main sound source as output, and performing model training through machine learning to obtain the sound positioning sub-model.
The sample group sound data includes sample group sound data collected by a plurality of sound collecting apparatuses. In the process of determining the sound data of the sample group, the position of the sample pig which emits the maximum decibel sound in the group breeding fence at the set sampling time point can be determined by the breeding personnel as the position of the main sound source of the sample.
Wherein the sound locator model may also be a convolutional neural network type model.
And obtaining the sound positioning sub-model through training of a plurality of groups of training samples. The training sample is data which takes the position of each sound pickup device for collecting the sample group sound data in the group culture bar, takes the sample group sound data collected at the set sampling time point as input, and takes the main sound source position of the sample as output.
The training of the sound positioning sub-model is realized through machine learning, positioning pigs corresponding to group sound data collected at each time point are separated through the sound positioning sub-model, individual sound data corresponding to each pig are determined, identification of abnormal pigs is realized by combining sign change information and the individual sound data, and accuracy is higher.
Further, on the basis of the above embodiments, the sound separation subunit is configured to:
acquiring a single-frame sample image obtained by performing infrared image acquisition and/or shooting on the group culture bar at the set sampling time point according to the sample group image set, and acquiring a main sound source position of a sample corresponding to group sound data acquired at the set sampling time point according to the sound positioning sub-model;
Judging whether pigs with the distance to the main sound source position of the sample being smaller than or equal to a preset distance threshold exist according to the positions of the pigs in the single-frame sample image, if so, taking the pigs closest to the main sound source position of the sample as positioning pigs, and taking the group sound data acquired at the set sampling time point as the group sound data corresponding to the positioning pigs;
and acquiring group sound data corresponding to the pig of any sample from the group sound data acquired at each sampling time point in a sampling time period, and denoising the group sound data corresponding to the pig of any sample to obtain sample individual sound data corresponding to the pig of any sample.
And the group sound data corresponding to the pig of any sample is subjected to noise reduction, so that the sound data corresponding to the maximum decibel in the group sound data corresponding to the pig of any sample is reduced, namely the interference generated by the sound data corresponding to the maximum decibel of background noise is reduced.
In the embodiment, the sound separation subunit is used for determining the individual sound data corresponding to each pig by combining the main sound source position corresponding to the group sound data acquired at each set sampling time point by the sound positioning subunit and the position corresponding to each sample pig in the group image set, so that the identification of the abnormal pig is realized by combining the feature change information and the individual sound data, and the accuracy is higher.
Further, on the basis of the foregoing embodiments, the process of establishing the fusion processing sub-model includes:
acquiring an image segmentation sub-model, a physical sign extraction sub-model and a sound positioning sub-model which are obtained by performing model training through machine learning, and an input sub-unit and a sound separation sub-unit;
inputting the sample group image set into the image segmentation sub-model, and acquiring sample sign change information extracted by the sign extraction sub-model for any sample pig according to the individual image set of any sample pig in the image segmentation sub-model; the sample group image set is obtained by carrying out infrared image acquisition and/or camera shooting on the pigs in the group raising fence;
inputting sample group sound data into the input subunit, inputting the sample group sound data collected at the set sampling time point into the sound localization sub-model by the input subunit, outputting sample individual sound data of any sample pig by the sound separation subunit according to a sample main sound source position corresponding to the group sound data collected at each sampling time point in the sampling time period output by the sound localization sub-model and a position of each pig at each sampling time point in a sample group image;
And taking the combination of the sample sign change information of any sample pig and the sample individual sound data as input, taking the result of whether any sample pig is an abnormal pig or not as output, and performing model training through machine learning to obtain the fusion processing sub-model.
Wherein, the fusion processing submodel can also be a convolutional neural network type model.
The training of the fusion processing submodel is realized by combining the input subunit and the sound separation subunit on the basis of the trained image segmentation submodel, the physical sign extraction submodel and the sound positioning submodel. The results of whether the sample pig is an abnormal pig or not are manually marked.
And performing model training through a plurality of groups of training samples and machine learning to obtain a fusion processing sub-model. The training sample comprises a sample which takes the combination of sample sign change information and sample individual sound data as input and takes the result of whether a sample pig is an abnormal pig or not as output.
In the embodiment, the training process of the fusion processing sub-model is realized through the image segmentation sub-model, the physical sign extraction sub-model, the sound positioning sub-model, the input sub-unit and the sound separation sub-unit. The fusion processing submodel combines the sign change information and the individual voice data to realize the identification of the abnormal pigs, and the identification accuracy of the abnormal pigs is higher.
The embodiment provides an abnormal pig identification system based on image and audio mixing, which comprises infrared cameras and/or video cameras arranged at different positions of a group raising fence, sound pickup equipment arranged at different positions of the group raising fence, and any one of the abnormal pig identification devices based on image and audio mixing;
the infrared camera is used for carrying out infrared image acquisition on the group breeding fence with the pigs in captivity to obtain a group image set comprising infrared images, and the obtained group image set is sent to the abnormal pig identification device based on image and audio mixing;
the camera is used for shooting the group breeding fence to obtain a group image set comprising shot images, and sending the obtained group image set to the abnormal pig identification device based on image and audio mixing;
the sound pickup equipment is used for collecting sound of the group raising fence to obtain group sound data, and sending the obtained group sound data to the abnormal pig identification device based on image and audio mixing.
The infrared cameras are arranged on the periphery and/or the top of the group-raising fence and used for collecting infrared images of the pigs in the group-raising fence in all directions. The cameras are also arranged on the periphery and/or the top of the group-raising fence and are used for collecting images of the pigs in the group-raising fence in an all-around mode. The sound pickup equipment comprises a plurality of sound pickup equipment, and the sound pickup equipment is arranged at different positions of the group breeding fence, for example, one sound pickup equipment is arranged at each of four directions of the south, the east and the north of the group breeding fence.
The embodiment provides an abnormal pig recognition system based on image and audio mixing, which comprises an infrared camera and/or a video camera, a sound pickup device and an abnormal pig recognition device based on image and audio mixing. The system acquires the group image set and the group voice data of the pigs in the group raising fence, and identifies whether each pig is an abnormal pig or not according to the group image set and the group voice data through the identification model in the identification module. The identification model comprises a fusion processing sub-model, and the fusion processing sub-model outputs the identification result of whether any pig is an abnormal pig or not according to the sign change information and the individual voice data of any pig determined by the group image set and the group voice data. Through the combination of the image and the sound, the accuracy of identifying the abnormal pig is improved, and the reliability of the detection result is high. Meanwhile, manual detection is not needed, the detection efficiency of abnormal pigs is improved, epidemic prevention and control are facilitated, and loss is reduced.
The embodiment provides an abnormal pig identification method based on image and audio mixing, which comprises the following steps:
s1: the acquisition module acquires a group image set obtained by carrying out infrared image acquisition and/or camera shooting on a group raising fence in which pigs are confined within a preset time period, and group sound data acquired by the pigs in the group raising fence within the preset time period;
S2: the identification module inputs the group image set and the group sound data into an identification model to obtain an identification result of whether each pig of the group raising fence output by the identification model is an abnormal pig or not;
the identification model comprises a fusion processing submodel, and the fusion processing submodel outputs the result of whether any pig is an abnormal pig or not according to the combination of the sign change information and the individual sound data of any pig; the sign change information of any pig is determined according to the individual image set corresponding to any pig in the group image set; the individual sound data of any pig is determined according to the group sound data and the group image set.
The method for identifying the abnormal pig based on the image and audio mixing provided by the embodiment is suitable for the device for identifying the abnormal pig based on the image and audio mixing provided by the embodiments, and is not described herein again.
The embodiment provides an abnormal pig identification method based on image and audio mixing, which is used for acquiring a group image set and a group voice data of pigs in a group fence, and identifying whether each pig is an abnormal pig or not according to the group image set and the group voice data through an identification model in an identification module. The identification model comprises a fusion processing sub-model, and the fusion processing sub-model outputs the identification result of whether any pig is an abnormal pig or not according to the sign change information and the individual voice data of any pig determined by the group image set and the group voice data. Through the combination of the image and the sound, the accuracy of identifying the abnormal pig is improved, and the reliability of the detection result is high. Meanwhile, manual detection is not needed, the detection efficiency of abnormal pigs is improved, epidemic prevention and control are facilitated, and loss is reduced.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. An abnormal pig identification device based on image and audio mixing is characterized by comprising an acquisition module and an identification module;
the acquisition module acquires a group image set obtained by carrying out infrared image acquisition and/or camera shooting on a group raising fence in which pigs are confined within a preset time period, and group sound data acquired by the pigs in the group raising fence within the preset time period;
the identification module inputs the group image set and the group sound data into an identification model to obtain an identification result of whether each pig of the group raising fence output by the identification model is an abnormal pig or not;
the identification model comprises a fusion processing submodel, and the fusion processing submodel outputs the result of whether any pig is an abnormal pig or not according to the combination of the sign change information and the individual sound data of any pig; the sign change information of any pig is determined according to the individual image set corresponding to any pig in the group image set; the individual sound data of any pig is determined according to the group sound data and the group image set.
2. The abnormal pig identification device based on image and audio mixing of claim 1, wherein the identification model further comprises an image segmentation sub-model, a physical sign extraction sub-model, an input sub-unit, a sound localization sub-model and a sound ionization unit:
the image segmentation sub-model performs image segmentation according to the group images to obtain an individual image set corresponding to each pig; for any pig, the sign extraction sub-model is used for extracting sign change information of any pig according to the individual image set corresponding to any pig;
the input subunit is used for acquiring group sound data acquired at any time point according to the group sound data and inputting the group sound data acquired at any time point into the sound positioning sub-model;
the sound positioning sub-model determines a main sound source position corresponding to the group sound data collected at any time point according to the position of pickup equipment collecting the group sound data in the group fence and the group sound data collected at any time point; the sound separation subunit outputs individual sound data corresponding to each pig according to the main sound source position corresponding to the group sound data collected at each time point output by the sound positioning submodel and the pig position of each pig at different time points determined according to the group image set;
The fusion processing submodel outputs a result whether any pig is an abnormal pig or not according to the combination of the sign change information output by the sign extraction submodel for any pig and the individual sound data corresponding to any pig output by the sound separation submodel;
and the position of the main sound source represents the position of the pig which emits the maximum decibel sound in the group-raising fence at any time point.
3. The abnormal pig recognition device based on image and audio mixing of claim 2, wherein the sound separation subunit is configured to:
acquiring a single-frame image obtained by acquiring and/or shooting an infrared image and/or a camera of the group fence at any time point according to the group image set, and acquiring a main sound source position corresponding to group sound data acquired at any time point according to the sound positioning sub-model;
judging whether pigs with the distance to the main sound source position smaller than or equal to a preset distance threshold exist according to the positions of the pigs in the single-frame image, if so, taking the pigs closest to the main sound source position as positioning pigs, and taking the group sound data acquired at any time point as the group sound data corresponding to the positioning pigs;
And acquiring group sound data corresponding to any pig from the group sound data acquired at each time point in the preset time period, and denoising the group sound data corresponding to any pig to obtain individual sound data corresponding to any pig.
4. The abnormal pig identification device based on image and audio mixing of claim 2, wherein the image segmentation submodel is established by the following steps:
acquiring a plurality of group image sets obtained by carrying out infrared image acquisition and/or camera shooting on pigs in a group raising fence in a sampling time period, taking the group image sets as sample group image sets, and segmenting a sample individual image set corresponding to each pig from the sample group image sets;
and taking the sample group image set as an input, taking the sample individual image set corresponding to each divided pig as an output, and performing model training through machine learning to obtain the image division sub-model.
5. The abnormal pig identification device based on image and audio mixing of claim 2, wherein the establishment process of the sign extraction submodel comprises:
obtaining a sample individual image set of any sample pig, wherein the sample individual image set comprises at least one of the following images: a first image subset continuously acquired during a first time period when any sample pig is in a cough state, a second image subset continuously acquired during a second time period when any sample pig is in a doze state, and an infrared image subset continuously acquired during a sampling time period;
Taking the sample individual image set as input, taking sample sign change information as output, and performing model training through machine learning to obtain the sign extraction sub-model; wherein, the sample sign change information comprises at least one of the following information: information corresponding to the first subset of images that a cough has occurred in the pig sample during the first time period, information corresponding to the second subset of images that the pig sample during the second time period is dozing, and information corresponding to a change in body temperature of the pig in the infrared subset of images;
wherein the second time period within the first time period is a sub-time period in a sampling time period for acquiring the sample group image set; the first image subset, the second image subset, and the infrared image subset are subsets of the sample individual image set.
6. The abnormal pig identification device based on image and audio mixing of claim 2, wherein the establishing process of the sound localization submodel comprises:
acquiring the position of each sound pickup device for acquiring group sound data of pigs in the group raising fence, and the group sound data acquired by each sound pickup device at a set sampling time point as sample group sound data; at the set sampling time point, the sample pig which sends the maximum decibel sound in the group-raising fence is only positioned at the position of the main sound source of the sample;
And taking the position of each sound pickup device for collecting the sample group sound data in the group culture bar, the sample group sound data collected at the set sampling time point as input, and taking the position of the sample main sound source as output, and performing model training through machine learning to obtain the sound positioning sub-model.
7. The abnormal pig identification device based on image and audio mixing of claim 6, wherein the sound separation subunit is configured to:
acquiring a single-frame sample image obtained by performing infrared image acquisition and/or shooting on the group culture bar at the set sampling time point according to the sample group image set, and acquiring a main sound source position of a sample corresponding to group sound data acquired at the set sampling time point according to the sound positioning sub-model;
judging whether pigs with the distance to the main sound source position of the sample being smaller than or equal to a preset distance threshold exist according to the positions of the pigs in the single-frame sample image, if so, taking the pigs closest to the main sound source position of the sample as positioning pigs, and taking the group sound data acquired at the set sampling time point as the group sound data corresponding to the positioning pigs;
And acquiring group sound data corresponding to the pig of any sample from the group sound data acquired at each sampling time point in a sampling time period, and denoising the group sound data corresponding to the pig of any sample to obtain sample individual sound data corresponding to the pig of any sample.
8. The apparatus for identifying abnormal pig based on image and audio mixing according to claim 7, wherein the process of establishing the fusion processing submodel comprises:
acquiring an image segmentation sub-model, a physical sign extraction sub-model and a sound positioning sub-model which are obtained by performing model training through machine learning, and an input sub-unit and a sound separation sub-unit;
inputting the sample group image set into the image segmentation sub-model, and acquiring sample sign change information extracted by the sign extraction sub-model for any sample pig according to the individual image set of any sample pig in the image segmentation sub-model; the sample group image set is obtained by carrying out infrared image acquisition and/or camera shooting on the pigs in the group raising fence;
inputting sample group sound data into the input subunit, inputting the sample group sound data collected at the set sampling time point into the sound localization sub-model by the input subunit, outputting sample individual sound data of any sample pig by the sound separation subunit according to a sample main sound source position corresponding to the group sound data collected at each sampling time point in the sampling time period output by the sound localization sub-model and a position of each pig at each sampling time point in a sample group image;
And taking the combination of the sample sign change information of any sample pig and the sample individual sound data as input, taking the result of whether any sample pig is an abnormal pig or not as output, and performing model training through machine learning to obtain the fusion processing sub-model.
9. An abnormal pig recognition system based on image and audio mixing, which is characterized by comprising infrared cameras and/or video cameras arranged at different positions of a group raising fence, sound pickup equipment arranged at different positions of the group raising fence, and the abnormal pig recognition device based on image and audio mixing of any one of claims 1 to 8;
the infrared camera is used for carrying out infrared image acquisition on the group breeding fence with the pigs in captivity to obtain a group image set comprising infrared images, and the obtained group image set is sent to the abnormal pig identification device based on image and audio mixing;
the camera is used for shooting the group breeding fence to obtain a group image set comprising shot images, and sending the obtained group image set to the abnormal pig identification device based on image and audio mixing;
the sound pickup equipment is used for collecting sound of the group raising fence to obtain group sound data, and sending the obtained group sound data to the abnormal pig identification device based on image and audio mixing.
10. An abnormal pig identification method based on image and audio mixing is characterized by comprising the following steps:
the acquisition module acquires a group image set obtained by carrying out infrared image acquisition and/or camera shooting on a group raising fence in which pigs are confined within a preset time period, and group sound data acquired by the pigs in the group raising fence within the preset time period;
the identification module inputs the group image set and the group sound data into an identification model to obtain an identification result of whether each pig of the group raising fence output by the identification model is an abnormal pig or not;
the identification model comprises a fusion processing submodel, and the fusion processing submodel outputs the result of whether any pig is an abnormal pig or not according to the combination of the sign change information and the individual sound data of any pig; the sign change information of any pig is determined according to the individual image set corresponding to any pig in the group image set; the individual sound data of any pig is determined according to the group sound data and the group image set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010600412.4A CN111860203B (en) | 2020-06-28 | 2020-06-28 | Abnormal pig identification device, system and method based on image and audio mixing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010600412.4A CN111860203B (en) | 2020-06-28 | 2020-06-28 | Abnormal pig identification device, system and method based on image and audio mixing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111860203A true CN111860203A (en) | 2020-10-30 |
| CN111860203B CN111860203B (en) | 2023-09-01 |
Family
ID=72988744
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010600412.4A Active CN111860203B (en) | 2020-06-28 | 2020-06-28 | Abnormal pig identification device, system and method based on image and audio mixing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111860203B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113537064A (en) * | 2021-07-16 | 2021-10-22 | 河南牧原智能科技有限公司 | Weak pig automatic detection marking method and system |
| CN113709378A (en) * | 2021-09-08 | 2021-11-26 | 联想(北京)有限公司 | Processing method and device, camera equipment and electronic system |
| CN113762219A (en) * | 2021-11-03 | 2021-12-07 | 恒林家居股份有限公司 | Method, system and storage medium for identifying people in mobile conference room |
| CN117423342A (en) * | 2023-10-27 | 2024-01-19 | 东北农业大学 | Pig abnormal state monitoring method and system based on edge calculation |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104064011A (en) * | 2014-07-02 | 2014-09-24 | 江苏大学 | Breeding pig delivery room monitoring system and method based on wireless transmission |
| CN107182830A (en) * | 2017-06-09 | 2017-09-22 | 中国农业科学院农业信息研究所 | A kind of sow farrowing bed monitoring warning device, method and system |
| CN109493874A (en) * | 2018-11-23 | 2019-03-19 | 东北农业大学 | A kind of live pig cough sound recognition methods based on convolutional neural networks |
| CN109637549A (en) * | 2018-12-13 | 2019-04-16 | 北京小龙潜行科技有限公司 | A kind of pair of pig carries out the method, apparatus and detection system of sound detection |
| CN109684953A (en) * | 2018-12-13 | 2019-04-26 | 北京小龙潜行科技有限公司 | The method and device of pig tracking is carried out based on target detection and particle filter algorithm |
| CN110598643A (en) * | 2019-09-16 | 2019-12-20 | 上海秒针网络科技有限公司 | Method and device for monitoring piglet compression |
-
2020
- 2020-06-28 CN CN202010600412.4A patent/CN111860203B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104064011A (en) * | 2014-07-02 | 2014-09-24 | 江苏大学 | Breeding pig delivery room monitoring system and method based on wireless transmission |
| CN107182830A (en) * | 2017-06-09 | 2017-09-22 | 中国农业科学院农业信息研究所 | A kind of sow farrowing bed monitoring warning device, method and system |
| CN109493874A (en) * | 2018-11-23 | 2019-03-19 | 东北农业大学 | A kind of live pig cough sound recognition methods based on convolutional neural networks |
| CN109637549A (en) * | 2018-12-13 | 2019-04-16 | 北京小龙潜行科技有限公司 | A kind of pair of pig carries out the method, apparatus and detection system of sound detection |
| CN109684953A (en) * | 2018-12-13 | 2019-04-26 | 北京小龙潜行科技有限公司 | The method and device of pig tracking is carried out based on target detection and particle filter algorithm |
| CN110598643A (en) * | 2019-09-16 | 2019-12-20 | 上海秒针网络科技有限公司 | Method and device for monitoring piglet compression |
Non-Patent Citations (2)
| Title |
|---|
| 张栖铭;袁瑞临;范凡;王峰;: "基于SVM算法的猪声音识别的研究", 电脑知识与技术, no. 10 * |
| 苍岩;罗顺元;乔玉龙;: "基于深层神经网络的猪声音分类", 农业工程学报, no. 09 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113537064A (en) * | 2021-07-16 | 2021-10-22 | 河南牧原智能科技有限公司 | Weak pig automatic detection marking method and system |
| CN113709378A (en) * | 2021-09-08 | 2021-11-26 | 联想(北京)有限公司 | Processing method and device, camera equipment and electronic system |
| CN113762219A (en) * | 2021-11-03 | 2021-12-07 | 恒林家居股份有限公司 | Method, system and storage medium for identifying people in mobile conference room |
| CN117423342A (en) * | 2023-10-27 | 2024-01-19 | 东北农业大学 | Pig abnormal state monitoring method and system based on edge calculation |
| CN117423342B (en) * | 2023-10-27 | 2024-06-07 | 东北农业大学 | Pig abnormal state monitoring method and system based on edge calculation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111860203B (en) | 2023-09-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111860203A (en) | Abnormal pig identification device, system and method based on image and audio mixing | |
| CN110705405B (en) | Target labeling method and device | |
| CN108090458B (en) | Human body falling detection method and device | |
| CN109299683B (en) | Security protection evaluation system based on face recognition and behavior big data | |
| ZA202300610B (en) | System and method for crop monitoring | |
| CN111968152B (en) | Dynamic identity recognition method and device | |
| CN111898581A (en) | Animal detection method, device, electronic equipment and readable storage medium | |
| CN110781711A (en) | Target object identification method and device, electronic equipment and storage medium | |
| CN110610127B (en) | Face recognition method and device, storage medium and electronic equipment | |
| KR102015945B1 (en) | Method for packaging learning images for atonomous vehicle and apparatus thereof | |
| WO2019119396A1 (en) | Facial expression recognition method and device | |
| CN112614109B (en) | Image quality evaluation method, apparatus and computer readable storage medium | |
| CN112287741A (en) | Image processing-based farming operation management method and device | |
| CN111325133A (en) | Image processing system based on artificial intelligence recognition | |
| CN116546023A (en) | Method and system for identifying violent behaviors of oil and gas operation area | |
| CN110929555B (en) | Face recognition method and electronic device using same | |
| CN112885014A (en) | Early warning method, device, system and computer readable storage medium | |
| CN115376210B (en) | Drowning behavior identification method, device, equipment and medium for preventing drowning in swimming pool | |
| WO2020080626A1 (en) | Method and apparatus for identifying animal species | |
| CN111582233A (en) | Data processing method, electronic device, and storage medium | |
| CN113553965B (en) | Person identity recognition method combining face recognition and human body recognition | |
| CN113255766B (en) | Image classification method, device, equipment and storage medium | |
| CN114283279A (en) | Travel card identification method applied to infectious disease prevention and control and infectious disease prevention and control method | |
| CN112580462A (en) | Feature point selection method, terminal and storage medium | |
| CN111160279A (en) | Method, apparatus, device and medium for generating target recognition model using small sample |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |