CN112070185A - Re-ID-based non-contact fever person tracking system and tracking method thereof - Google Patents
Re-ID-based non-contact fever person tracking system and tracking method thereof Download PDFInfo
- Publication number
- CN112070185A CN112070185A CN202010810436.2A CN202010810436A CN112070185A CN 112070185 A CN112070185 A CN 112070185A CN 202010810436 A CN202010810436 A CN 202010810436A CN 112070185 A CN112070185 A CN 112070185A
- Authority
- CN
- China
- Prior art keywords
- retrieval
- body temperature
- person
- sample
- neural network
- 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
- 238000000034 method Methods 0.000 title claims abstract description 33
- 206010037660 Pyrexia Diseases 0.000 title claims abstract description 30
- 230000036760 body temperature Effects 0.000 claims abstract description 63
- 230000002159 abnormal effect Effects 0.000 claims abstract description 51
- 238000003062 neural network model Methods 0.000 claims abstract description 41
- 238000012544 monitoring process Methods 0.000 claims abstract description 29
- 238000012163 sequencing technique Methods 0.000 claims abstract description 23
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000004891 communication Methods 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 19
- 239000011159 matrix material Substances 0.000 claims description 15
- 239000013598 vector Substances 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 5
- 238000001514 detection method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000009529 body temperature measurement Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000001681 protective effect Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 208000003322 Coinfection Diseases 0.000 description 1
- 208000035473 Communicable disease Diseases 0.000 description 1
- 241000711573 Coronaviridae Species 0.000 description 1
- 241000700605 Viruses Species 0.000 description 1
- 239000000645 desinfectant Substances 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 239000012567 medical material Substances 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K17/00—Methods or arrangements for effecting co-operative working between equipments covered by two or more of main groups G06K1/00 - G06K15/00, e.g. automatic card files incorporating conveying and reading operations
- G06K17/0022—Methods or arrangements for effecting co-operative working between equipments covered by two or more of main groups G06K1/00 - G06K15/00, e.g. automatic card files incorporating conveying and reading operations arrangements or provisious for transferring data to distant stations, e.g. from a sensing device
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/01—Measuring temperature of body parts ; Diagnostic temperature sensing, e.g. for malignant or inflamed tissue
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/74—Details of notification to user or communication with user or patient ; user input means
- A61B5/746—Alarms related to a physiological condition, e.g. details of setting alarm thresholds or avoiding false alarms
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
- G01J5/0022—Radiation pyrometry, e.g. infrared or optical thermometry for sensing the radiation of moving bodies
- G01J5/0025—Living bodies
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/53—Querying
- G06F16/532—Query formulation, e.g. graphical querying
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
- G01J5/00—Radiation pyrometry, e.g. infrared or optical thermometry
- G01J2005/0077—Imaging
Abstract
The invention discloses a Re-ID-based non-contact type fever personnel tracking system and a tracking method thereof, wherein the Re-ID-based non-contact type fever personnel tracking system comprises a target acquisition front end, an infrared thermal imager and at least two cameras electrically connected with the infrared thermal imager; at least two surveillance cameras; the main control end is respectively in communication connection with the target acquisition front end and the monitoring camera group; the Re-ID-based non-contact fever person tracking method comprises the steps of finding out a person with abnormal body temperature and a portrait of the person with abnormal body temperature; acquiring a shooting picture of a monitoring camera; processing the portrait and the shot picture by using a yolov3 model to obtain a retrieval image and a retrieval sample; preprocessing the retrieval image and the retrieval sample, inputting the preprocessed retrieval image and retrieval sample into a re-recognition neural network model, obtaining sequencing data and analyzing the trace of the person with abnormal body temperature; the embodiment of the invention can detect the body temperature of a human body in a non-contact manner, find a person with abnormal body temperature and then track the trace of the person with abnormal body temperature.
Description
Technical Field
The invention relates to the technical field of pedestrian tracking, in particular to a Re-ID-based non-contact fever person tracking system and a tracking method thereof.
Background
When some infectious diseases such as novel coronavirus abuse occur frequently, in order to control the epidemic situation, body temperature detection needs to be carried out on each person, the whereabouts of each person are surveyed, infected persons and persons who contact the infected persons can be found out timely, isolation treatment is carried out on the infected persons and the persons who contact the infected persons, and diffusion of the viruses is controlled timely; in the mode, workers need to be arranged at each detection point, each worker needs to be equipped with disposable consumables such as a mask, a disinfectant, a protective suit and the like and detection tools such as a body temperature gun and the like, much manpower and material resources need to be consumed, and the difficulty degree of epidemic prevention and control is emphasized under the condition of shortage of materials; moreover, if the track of the person with abnormal body temperature is to be found, the person to be measured at each inspection point needs to be manually checked for personal information, and this tracking method is very inefficient.
Disclosure of Invention
The present invention is directed to solving at least one of the problems of the prior art. Therefore, the invention provides a Re-ID-based non-contact fever person tracking system which can detect the body temperature of a human body in a non-contact mode, find a person with abnormal body temperature and track the track of the person with abnormal body temperature.
The invention also provides a non-contact fever staff tracking method based on the Re-ID based non-contact fever staff tracking system.
According to the embodiment of the first aspect of the invention, the Re-ID-based non-contact fever staff tracking system comprises:
the target acquisition front end comprises an infrared thermal imager and at least two cameras, wherein the infrared thermal imager is electrically connected with the cameras;
the monitoring camera group comprises at least two monitoring cameras;
the master control end is in communication connection with the target acquisition front end and the monitoring camera group respectively and can store and process information sent by the target acquisition front end and the monitoring camera group.
The Re-ID-based non-contact fever staff tracking system has at least the following beneficial effects: the method comprises the steps that a target acquisition front end detects and obtains a portrait of a person with abnormal body temperature, the portrait is sent to a main control end, a monitoring camera near the target acquisition front end sends a shot picture to the main control end, the main control end processes the portrait of the person with abnormal body temperature and sample portraits in a plurality of shot pictures based on Re-ID technology, the sample portraits with the highest similarity with the person with abnormal body temperature in the plurality of shot pictures are obtained after processing, and the track of the person with abnormal body temperature is obtained according to time and position information of the sample portraits.
According to some embodiments of the present invention, the master control end includes a controller, a computer-readable storage medium, and a processor, the controller is respectively in communication connection with the target acquisition front end and the monitoring camera group, the computer-readable storage medium can store a computer program and data sent by the controller, and the processor can run the computer program.
According to a second aspect of the invention, the Re-ID-based non-contact fever staff tracking method comprises the following steps:
finding the abnormal body temperature person and acquiring a portrait of the abnormal body temperature person;
acquiring a shooting picture of the monitoring camera;
respectively processing the portrait and the shot picture by using a yolov3 model to respectively obtain a retrieval map and a retrieval map library, wherein the retrieval map library comprises at least two retrieval samples;
preprocessing the retrieval map and the retrieval sample;
inputting the retrieval graph and the retrieval sample into a re-recognition neural network model to obtain sequencing data;
and obtaining the track of the person with abnormal body temperature according to the sequencing data.
The Re-ID-based non-contact fever staff tracking method has at least the following beneficial effects: the method comprises the steps of respectively acquiring a portrait and a shot picture of a person with abnormal body temperature by using a target acquisition front end and a monitoring camera group, then respectively processing the portrait and the shot picture through a yolov3 model, preprocessing and re-identifying a neural network model to obtain sequencing data, and finally obtaining the track of the person with abnormal body temperature through the sequencing data, so that manual temperature measurement is not needed, manpower and material resources are saved, and meanwhile, the tracking efficiency can be greatly improved.
According to some embodiments of the invention, the discovering the abnormal body temperature person and acquiring the portrait of the abnormal body temperature person comprises:
the infrared thermal imager measures the body temperature of the measured person;
if the body temperature is larger than a preset value, the infrared thermal imager sends alarm information to the camera head;
the camera shoots the person with abnormal body temperature to obtain the portrait.
According to some embodiments of the present invention, said processing said portrait and said captured picture, respectively, using yolov3 model, resulting in a search map and a search gallery, respectively, said search gallery comprising at least two search samples, comprises the following steps before:
the yolov3 model was trained to recognize a person.
According to some embodiments of the invention, the preprocessing the search graph and the search sample comprises: and performing parameter rectification, ISP optimization and noise removal on the retrieval graph and the retrieval sample.
According to some embodiments of the invention, the inputting the search graph and the search sample into a re-recognition neural network model, and the obtaining the ranking data comprises:
extracting first characteristic information of the retrieval map and second characteristic information of the retrieval sample;
converting the retrieval graph into a first matrix vector according to the first characteristic information, and converting the retrieval sample into a second matrix vector according to the second characteristic information;
and respectively calculating the distance between the first matrix vector and the second matrix vector of each retrieval sample, and sequencing each retrieval sample according to the distance to obtain the sequencing data.
According to some embodiments of the invention, before inputting the search graph and the search sample into the re-recognition neural network model to obtain the ranking data, the method comprises the following steps:
building an original re-recognition neural network model;
importing training data into the original re-recognition neural network model to obtain training sequencing data;
calculating the first hit rate and the average precision according to the training sequencing data;
and if the head hit rate is greater than a preset head hit rate and the average precision is greater than a preset average precision, obtaining the re-recognition neural network model.
According to some embodiments of the invention, the constructing the original re-recognition neural network model comprises:
using Resnet50 as the underlying backbone network;
and adding models of learning global information branches and concerned local detail information branches into the basic backbone network to obtain the original re-recognition neural network model.
According to some embodiments of the invention, the training data comprises a Market1501 data set and a DukeMTMC-reid data set.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
FIG. 1 is a schematic structural diagram of a Re-ID-based non-contact fever staff tracking system according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a Re-ID-based non-contact fever tracking method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a retrieval diagram and a retrieval sample input Re-recognition neural network model of the Re-ID-based non-contact fever staff tracking method shown in FIG. 2, and a flow chart of sequencing data is obtained;
FIG. 4 is a schematic flow chart of a Re-identification neural network model obtained by the Re-ID-based non-contact fever staff tracking method shown in FIG. 2;
reference numerals:
a target acquisition front end 100, an infrared thermal imager 110, a camera 120,
A monitoring camera group 200, a monitoring camera 210,
A main control end 300, a controller 310, a computer readable storage medium 320, and a processor 330.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
In the description of the present invention, the meaning of a plurality of means is one or more, the meaning of a plurality of means is two or more, and larger, smaller, larger, etc. are understood as excluding the number, and larger, smaller, inner, etc. are understood as including the number.
In the description of the present invention, unless otherwise explicitly defined, terms such as arrangement, connection and the like should be broadly construed, and those skilled in the art can reasonably determine the specific meanings of the above terms in the present invention in combination with the detailed contents of the technical solutions.
Referring to fig. 1, in the present embodiment, a non-contact fever personnel tracking system based on Re-ID includes: the target acquisition front end 100 comprises an infrared thermal imager 110 and at least two cameras 120, wherein the infrared thermal imager 110 is electrically connected with the cameras 120; the surveillance camera group 200 comprises at least two surveillance cameras 210; the main control terminal 300 is in communication connection with the target acquisition front end 100 and the monitoring camera group 200 respectively, and can store and process information sent by the target acquisition front end 100 and the monitoring camera group 200; firstly, placing the target acquisition front end 100 at a plurality of places, carrying out body temperature detection on each pedestrian passing by the infrared thermal imager 110, if the body temperature of the pedestrian exceeds a preset value, determining that the pedestrian is abnormal in body temperature, sending an alarm signal to the camera 120 by the infrared thermal imager 110, shooting the abnormal in body temperature after receiving the alarm signal by the camera 120, respectively acquiring a plurality of photos of the abnormal in body temperature by the cameras 120 with at least two non-overlapping visual fields and sending the photos to the main control end 300, calling and analyzing a shooting picture of the monitoring camera 210 by the main control end 300, extracting a sample portrait in the shooting picture, then comparing the portrait of the abnormal in body temperature with each sample portrait to obtain a sample portrait most similar to the portrait of the abnormal in body temperature, determining the most similar sample portrait as the abnormal in body temperature shot by the monitoring camera, and then according to time and place information of each most similar sample portrait, the track of the person with abnormal body temperature is obtained, so that the tracking efficiency of the staff is greatly improved; in some embodiments, the master control end 300 passes through the data line with camera 120 and surveillance camera group 200 respectively, communication connection is realized to wireless network or bluetooth, in this embodiment, the master control end passes through wireless network connection with camera 120 and surveillance camera group 200 respectively, camera 120 and surveillance camera group 200 pass through wireless network with the portrait of shooing and shoot the long-range sending of picture for the master control end, thereby the staff can remotely obtain the portrait of the abnormal person of body temperature, need not to contact with the personnel being tested, effectively avoid by the risk of secondary infection, need not to arrange the staff at every check point simultaneously, greatly save the consumption of medical materials such as gauze mask, protective clothing.
Referring to fig. 1, in some embodiments, the main control terminal 300 includes a controller 310, a computer-readable storage medium 320, and a processor 330, the controller 310 is respectively in communication connection with the target acquisition front end 100 and the monitoring camera group 200, the computer-readable storage medium 320 can store a computer program and data sent by the controller 310, the processor 330 can run the computer program, so that the target acquisition front end 100 and the monitoring camera group 200 respectively send the portrait of the abnormal body temperature person and the shot picture to the controller 310, the controller 310 sends the portrait of the abnormal body temperature person and the shot picture to the processor 330 and sends them to the computer-readable storage medium 320 for storage, and the processor 330 runs the computer program, and analyzes the portrait of the abnormal body temperature person and the shot picture to obtain the whereabouts of the abnormal body temperature person.
Referring to fig. 1 to 2, in the present embodiment, a non-contact fever staff tracking method based on Re-ID based non-contact fever staff tracking system includes: s110, finding out a person with abnormal body temperature, and acquiring a portrait of the person with abnormal body temperature; s120, acquiring a shooting picture of the monitoring camera; s210, respectively processing the portrait and the shot picture by using a yolov3 model to respectively obtain a retrieval image and a retrieval image library, wherein the retrieval image library comprises at least two retrieval samples; s220, preprocessing the retrieval image and the retrieval sample; s400, inputting the retrieval image and the retrieval sample into a re-recognition neural network model to obtain sequencing data; s500, obtaining the track of the person with abnormal body temperature according to the sequencing data.
Firstly, the target acquisition front end 100 detects the body temperature of a pedestrian, images of a plurality of abnormal body temperature persons are respectively shot from the front and side angles and sent to the main control end 300, a yolov3 model can identify the pedestrian from one picture and divide the pedestrian by using a rectangular frame, the main control end 300 processes the images from the target acquisition front end 100 by using a yolov3 model to obtain retrieval pictures of the abnormal body temperature persons from different angles of appearance to disappearance in the visual field of the camera 120, the main control end 300 calls the shot picture of the monitoring camera group 200 and inputs the shot picture into a yolov3 model, the yolov3 model captures the pedestrian in the shot picture by using the rectangular frame to obtain a plurality of retrieval samples of the pedestrians appearing in the shot picture, and the retrieval samples form a retrieval picture library.
Then, each retrieval sample and each retrieval graph are preprocessed to be in accordance with the unified input standard of the re-recognition neural network model, the preprocessed retrieval samples and the preprocessed retrieval graphs are input into the trained re-recognition neural network model, the re-recognition neural network model respectively calculates Euclidean distances between the retrieval graphs and each retrieval sample, the smaller the Euclidean distances are, the higher the similarity between the retrieval samples and the retrieval graphs is, otherwise, the lower the similarity is, each retrieval sample is sequenced according to the Euclidean distances to obtain sequencing data, the sequencing data is analyzed to obtain a plurality of retrieval samples with the highest similarity to the retrieval graphs, the retrieval samples with the highest similarity to the retrieval graphs can be regarded as images of abnormal body temperature persons under the monitoring camera, and the tracks of the abnormal body temperature persons can be analyzed and obtained according to the time and place information of the retrieval samples with the highest similarity to each monitoring camera.
Referring to fig. 1-2, in some embodiments, the discovering the abnormal temperature subject and acquiring the portrait of the abnormal temperature subject at S110 includes: the infrared thermal imager 110 measures the body temperature of the measured person; if the body temperature is greater than the preset value, the infrared thermal imager 110 sends alarm information to the camera 120; the camera 120 shoots the abnormal body temperature person to obtain a portrait; every body temperature check point has all been placed infrared thermal imaging and has been had 110, and infrared thermal imager 110 is to each pedestrian's temperature measurement of process, and when detecting the pedestrian that the body temperature is unusual, infrared thermal imager 110 sends alarm information to camera 120, and camera 120 begins to shoot the portrait to the unusual person of body temperature after receiving alarm information to camera 120 need not to keep always shooting the state, avoids shooing unnecessary image, saves electric quantity and storage space.
Referring to fig. 2, in some embodiments, the step S210 of processing the portrait and the shot picture respectively by using the yolov3 model to obtain a search map and a search gallery respectively, wherein the search gallery comprises at least two search samples and the steps of: training the yolov3 model to identify people; the yolov3 model can identify a specific object according to the requirements of a user, so before processing a portrait and taking a picture by using the yolov3 model, the yolov3 model needs to be trained to be output as a human, namely, a person can be identified from an image and segmented by using a rectangular frame.
Referring to fig. 2, in some embodiments, the preprocessing of the retrieval map and the retrieval sample by S220 includes: the method comprises the steps of performing parameter correction, ISP optimization and noise removal on a retrieval image and retrieval samples, wherein each retrieval sample is from different monitoring cameras, each retrieval sample and each retrieval image are obtained by shooting through different cameras, and the shooting environments and the shooting parameters of the different cameras are different, so that the illumination intensity, the noise and other information of each retrieval sample and each retrieval image are different, which affects the recognition accuracy of a re-recognition neural network model on the image, and therefore the retrieval samples and the retrieval images need to be preprocessed and subjected to parameter correction, ISP optimization and noise removal.
Referring to fig. 3, in some embodiments, the step S400 of inputting the search graph and the search sample into the re-recognition neural network model, and the obtaining the ranking data comprises: s410, extracting first characteristic information of a retrieval image and second characteristic information of a retrieval sample; s420, converting the retrieval image into a first matrix vector according to the first characteristic information, and converting the retrieval sample into a second matrix vector according to the second characteristic information; s430 respectively calculating the distance between the first matrix vector and the second matrix vector of each retrieval sample, and S440 sorting each retrieval sample according to the distance to obtain sorting data; each retrieval sample represents a pedestrian shot by a monitoring camera, the re-identification neural network model firstly extracts the characteristic information of the retrieval image and each retrieval sample respectively, then converts the retrieval image and each retrieval sample into matrix vectors respectively according to the characteristic information, then calculates the Euclidean distance between the matrix vector of the retrieval image and the matrix vector of each retrieval sample respectively, the smaller the Euclidean distance is, the more similar the retrieval sample and the retrieval image is, the more similar the pedestrian represented by the retrieval sample and the abnormal body temperature person is, if the Euclidean distance between the retrieval sample and the retrieval image is smaller than a preset value, the pedestrian represented by the retrieval sample is regarded as the abnormal body temperature person, and the time and place information of each retrieval sample with the Euclidean distance to the retrieval image smaller than the preset value is analyzed, so that the track of the abnormal body temperature person can be obtained.
Referring to fig. 4, in some embodiments, the step S400 of inputting the search graph and the search sample into the re-recognition neural network model comprises the following steps before obtaining the ranking data: building an original re-recognition neural network model; s330, importing training data into an original re-recognition neural network model to obtain training sequencing data; s340, calculating the first hit rate and the average precision according to the training sequencing data; s350, if the first hit rate is larger than the preset first hit rate and the average precision is larger than the preset average precision, S360 obtaining a re-recognition neural network model; if the first hit rate is not greater than the preset first hit rate or the average precision is not greater than the preset average precision, returning to step S330, importing the training data into the original re-recognition neural network model to obtain training sequencing data, and re-training the original re-recognition neural network model; the first hit rate and the average precision are parameters commonly used for evaluating the re-recognition neural network model, and the first hit rate and the average precision of the original re-recognition neural network model are both larger than a preset value through continuous training, so that the output of the re-recognition neural network model can be more reliable.
Referring to FIG. 4, in some embodiments, constructing the original re-recognition neural network model includes: s310 uses Resnet50 as the underlying backbone network; s320, adding models of learning global information branches and paying attention to local detail information branches into the basic backbone network to obtain an original re-recognition neural network model; resnet50 is a basic backbone network disclosed in the technical field of re-identification, a model for learning a global information branch and a local detail information branch is added into a Resnet50 model, and parameter initialization is performed after the model is called: num _ classes is set as datamanager, num _ train _ pid, optimization algorithm is set as adam, learning rate is set as 0.0003, lr _ scheduler is set as single _ step, and stepsize is set as 20, so that the original re-recognition neural network model required by the application can be obtained.
Referring to fig. 4, in some embodiments, the training data includes a Market1501 data set and a DukeMT MC-reid data set; the Market1501 data set and the DukeMTMC-reid data set are training data sets commonly used in the field of pedestrian re-recognition, and the re-recognition neural network model is trained by the training data sets, so that the original neural network model has a re-recognition function.
The embodiments of the present invention have been described in detail with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the gist of the present invention.
Claims (10)
1. A Re-ID based contactless fever personnel tracking system, comprising:
the target acquisition front end comprises an infrared thermal imager and at least two cameras, wherein the infrared thermal imager is electrically connected with the cameras;
the monitoring camera group comprises at least two monitoring cameras;
the master control end is in communication connection with the target acquisition front end and the monitoring camera group respectively and can store and process information sent by the target acquisition front end and the monitoring camera group.
2. A non-contact fever personnel tracking system based on Re-ID as claimed in claim 1, wherein the main control end comprises a controller, a computer readable storage medium and a processor, the controller is connected to the target collection front end and the surveillance camera group respectively, the computer readable storage medium can store a computer program and data sent by the controller, and the processor can run the computer program.
3. The tracking method applied to the Re-ID-based non-contact fever personnel tracking system is characterized by comprising the following steps:
identifying the abnormal body temperature person and acquiring a portrait of the abnormal body temperature person;
acquiring a shooting picture of the monitoring camera;
respectively processing the portrait and the shot picture by using a yolov3 model to respectively obtain a retrieval map and a retrieval map library, wherein the retrieval map library comprises at least two retrieval samples;
preprocessing the retrieval map and the retrieval sample;
inputting the retrieval graph and the retrieval sample into a re-recognition neural network model to obtain sequencing data;
and obtaining the track of the person with abnormal body temperature according to the sequencing data.
4. The Re-ID-based non-contact fever personnel tracking method as defined in claim 3, wherein the discovering the abnormal temperature person and obtaining the portrait of the abnormal temperature person comprises:
measuring the body temperature of the measured person based on the infrared thermal imager;
if the body temperature is larger than a preset value, the infrared thermal imager sends alarm information to the camera head;
and controlling/shooting the abnormal body temperature person based on the camera to obtain the portrait.
5. The Re-ID-based non-contact fever personnel tracking method as claimed in claim 3, wherein said using yolov3 model to process said portrait and said photographed picture respectively, and obtaining a search map and a search gallery respectively, wherein before said search gallery comprises at least two search samples, the method comprises the following steps:
the yolov3 model was trained to recognize a person.
6. The Re-ID-based non-contact fever personnel tracking method according to claim 3, wherein the pre-processing the search graph and the search sample comprises: and performing parameter rectification, ISP optimization and noise removal on the retrieval graph and the retrieval sample.
7. The method for contactless fever personnel tracking based on Re-ID as claimed in claim 3, wherein the inputting the search graph and the search sample into the Re-recognition neural network model to obtain the ranking data comprises:
extracting first characteristic information of the retrieval map and second characteristic information of the retrieval sample;
converting the retrieval graph into a first matrix vector according to the first characteristic information, and converting the retrieval sample into a second matrix vector according to the second characteristic information;
and respectively calculating the distance between the first matrix vector and the second matrix vector of each retrieval sample, and sequencing each retrieval sample according to the distance to obtain the sequencing data.
8. The Re-ID-based non-contact fever personnel tracking method according to claim 3, wherein the step of inputting the search graph and the search sample into a Re-recognition neural network model to obtain the ranking data comprises the following steps:
building an original re-recognition neural network model;
importing training data into the original re-recognition neural network model to obtain training sequencing data;
calculating the first hit rate and the average precision according to the training sequencing data;
and if the head hit rate is greater than a preset head hit rate and the average precision is greater than a preset average precision, obtaining the re-recognition neural network model.
9. The Re-ID-based non-contact fever personnel tracking method according to claim 8, wherein the constructing of the original Re-recognition neural network model comprises:
using Resnet50 as the underlying backbone network;
and adding models of learning global information branches and concerned local detail information branches into the basic backbone network to obtain the original re-recognition neural network model.
10. The method for contactless fever personnel tracking based on Re-ID as claimed in claim 8, wherein the training data comprises Market1501 data set and DukeMTMC-reid data set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010810436.2A CN112070185A (en) | 2020-08-13 | 2020-08-13 | Re-ID-based non-contact fever person tracking system and tracking method thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010810436.2A CN112070185A (en) | 2020-08-13 | 2020-08-13 | Re-ID-based non-contact fever person tracking system and tracking method thereof |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112070185A true CN112070185A (en) | 2020-12-11 |
Family
ID=73661567
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010810436.2A Pending CN112070185A (en) | 2020-08-13 | 2020-08-13 | Re-ID-based non-contact fever person tracking system and tracking method thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112070185A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112817342A (en) * | 2020-12-23 | 2021-05-18 | 上海工程技术大学 | Speaker tracking system integrating human body detection technology and wireless sensing technology and application |
US11467034B2 (en) * | 2020-06-30 | 2022-10-11 | Ilooda Co., Ltd. | Temperature measuring device for tracked subject target region |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109344787A (en) * | 2018-10-15 | 2019-02-15 | 浙江工业大学 | A kind of specific objective tracking identified again based on recognition of face and pedestrian |
CN109934104A (en) * | 2019-01-29 | 2019-06-25 | 武汉烽火众智数字技术有限责任公司 | The pedestrian retrieval method and system across camera lens identified again based on pedestrian |
CN110263697A (en) * | 2019-06-17 | 2019-09-20 | 哈尔滨工业大学(深圳) | Pedestrian based on unsupervised learning recognition methods, device and medium again |
CN111341461A (en) * | 2020-02-24 | 2020-06-26 | 中国船舶重工集团公司第七一七研究所 | Infrared epidemic situation intelligent comprehensive detection and identification system |
CN111414854A (en) * | 2020-03-19 | 2020-07-14 | 银河水滴科技(北京)有限公司 | Target crowd trajectory tracking method and device |
CN111426388A (en) * | 2020-03-31 | 2020-07-17 | 高新兴科技集团股份有限公司 | Personnel body temperature measuring method, system, computer storage medium and electronic equipment |
-
2020
- 2020-08-13 CN CN202010810436.2A patent/CN112070185A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109344787A (en) * | 2018-10-15 | 2019-02-15 | 浙江工业大学 | A kind of specific objective tracking identified again based on recognition of face and pedestrian |
CN109934104A (en) * | 2019-01-29 | 2019-06-25 | 武汉烽火众智数字技术有限责任公司 | The pedestrian retrieval method and system across camera lens identified again based on pedestrian |
CN110263697A (en) * | 2019-06-17 | 2019-09-20 | 哈尔滨工业大学(深圳) | Pedestrian based on unsupervised learning recognition methods, device and medium again |
CN111341461A (en) * | 2020-02-24 | 2020-06-26 | 中国船舶重工集团公司第七一七研究所 | Infrared epidemic situation intelligent comprehensive detection and identification system |
CN111414854A (en) * | 2020-03-19 | 2020-07-14 | 银河水滴科技(北京)有限公司 | Target crowd trajectory tracking method and device |
CN111426388A (en) * | 2020-03-31 | 2020-07-17 | 高新兴科技集团股份有限公司 | Personnel body temperature measuring method, system, computer storage medium and electronic equipment |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11467034B2 (en) * | 2020-06-30 | 2022-10-11 | Ilooda Co., Ltd. | Temperature measuring device for tracked subject target region |
CN112817342A (en) * | 2020-12-23 | 2021-05-18 | 上海工程技术大学 | Speaker tracking system integrating human body detection technology and wireless sensing technology and application |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR100886557B1 (en) | System and method for face recognition based on adaptive learning | |
CN106886216B (en) | Robot automatic tracking method and system based on RGBD face detection | |
JP5339476B2 (en) | Image processing system, fever tracking method, image processing apparatus, control method thereof, and control program | |
Chen et al. | A fall detection system based on infrared array sensors with tracking capability for the elderly at home | |
CN102859554B (en) | Collating device | |
JP5106356B2 (en) | Image monitoring device | |
JPWO2019038965A1 (en) | Store device, store management method, program | |
CN110956118B (en) | Target object detection method and device, storage medium and electronic device | |
CN109543607A (en) | Object abnormal state detection method, system, monitor system and storage medium | |
Kaixuan et al. | Target detection method for moving cows based on background subtraction | |
CN112070185A (en) | Re-ID-based non-contact fever person tracking system and tracking method thereof | |
JP6503079B2 (en) | Specific person detection system, specific person detection method and detection device | |
CN110728810B (en) | Distributed target monitoring system and method | |
CN110674680A (en) | Living body identification method, living body identification device and storage medium | |
JP6616906B1 (en) | Detection device and detection system for defective photographing data | |
CN111091057A (en) | Information processing method and device and computer readable storage medium | |
CN113688794A (en) | Identity recognition method and device, electronic equipment and computer readable storage medium | |
JP5758165B2 (en) | Article detection device and stationary person detection device | |
CN104392201A (en) | Human fall identification method based on omnidirectional visual sense | |
De Silva | Audiovisual sensing of human movements for home-care and security in a smart environment | |
CN114463779A (en) | Smoking identification method, device, equipment and storage medium | |
CN109982051A (en) | Monitoring camera method and monitoring camera with animal identification function | |
JP6893812B2 (en) | Object detector | |
KR101441285B1 (en) | Multi-body Detection Method based on a NCCAH(Normalized Cross-Correlation of Average Histogram) And Electronic Device supporting the same | |
TWI777689B (en) | Method of object identification and temperature measurement |
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 |