CN114170573A - School library seat occupation detection method based on YOLO v5 - Google Patents
School library seat occupation detection method based on YOLO v5 Download PDFInfo
- Publication number
- CN114170573A CN114170573A CN202111545932.0A CN202111545932A CN114170573A CN 114170573 A CN114170573 A CN 114170573A CN 202111545932 A CN202111545932 A CN 202111545932A CN 114170573 A CN114170573 A CN 114170573A
- Authority
- CN
- China
- Prior art keywords
- yolo
- seat
- person
- algorithm
- chair
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 26
- 238000012544 monitoring process Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 7
- 238000000034 method Methods 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 5
- 238000013135 deep learning Methods 0.000 claims description 5
- 238000003064 k means clustering Methods 0.000 claims description 4
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 2
- 230000002068 genetic effect Effects 0.000 claims description 2
- 238000003062 neural network model Methods 0.000 claims description 2
- 230000016273 neuron death Effects 0.000 claims description 2
- 238000011176 pooling Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 238000012795 verification Methods 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
- Alarm Systems (AREA)
Abstract
The invention relates to a campus library seat occupation detection method based on YOLO v 5. The camera and the YOLO V5 target detection algorithm monitor the vacant seat condition of the library in real time, for example, if someone on the seat can judge that someone on the seat, if no person is detected on the seat, it needs to be judged whether the seat occupying action is malicious or not at the moment, the lamp on the desk can be lighted to red, if no person is detected on the seat, the seat occupying action is considered, the timing is started at the moment, if the seat occupying time is more than 30 minutes, the seat occupying action is considered malicious at the moment, the lamp on the desk can be changed into green, the student without the seat can sit down, and the last person on the seat leaves for a long time or other students occupy a plurality of positions by one person. What is needed is to detect whether a person is on the seat in the first step, detect whether a large seat occupying object such as a book or a bag is on the table if no person is on the table, and calculate the placing time of the seat occupying object. Over 30 minutes this is considered a malicious seating occupation and other classmates can sit down with confidence.
Description
Technical Field
The invention relates to a campus library seat occupation detection method based on YOLO v 5.
Background
In recent years, with the increase of the number of examinees and the number of examinee public institutions, the number of seats in a library of a school is limited, so that the requirement that all students have seats, many students and friends help the seats is not met, a plurality of books or schoolbag seats are placed on the seats, and possibly no people are in the seat in the morning. Even if the students have disputes due to seat occupation, a device is needed to detect the seat occupation behavior, and the judgment by a machine can reduce the disputes and greatly improve the utilization rate of the seat, thereby preventing the vicious seat occupation.
Disclosure of Invention
The invention aims to provide a campus library occupancy detection method based on YOLO V5, and provides a monitoring system which utilizes a YOLO V5 detection model and a K-means clustering algorithm capable of obtaining a more reasonable anchor frame, quickly searches key points and identifies desk chairs, people, books, bags and other objects.
In order to achieve the purpose, the technical scheme of the invention is as follows: a campus library occupation detection method based on YOLO v5 adopts a YOLO 5 detection method fused with K-means to obtain a more reasonable anchor frame algorithm and improve the accuracy of classification target detection, and the method is concretely implemented by the following steps:
step S1, obtaining scene images from the monitoring videos of the monitoring areas of the cameras in the library, and detecting tables and chairs in the images according to a deep learning target detection algorithm framework YOLO 5;
s2, obtaining more reasonable 9 anchors frames needed by YOLO V5 by using a K-means clustering algorithm;
step S3, after the chair and the person are detected, the IOU ratio of the coincidence degree between the images of the chair and the person is calculated to judge whether the chair is occupied or not;
step S4, judging whether the distance is smaller than a threshold value, if so, judging that a person is on the seat, lighting a red light, and if not, continuing to detect the book and the bag;
step S5, if no book or bag is detected, then it is determined that no person is present and the position is sitting, and a green light is turned on;
and step S6, if the book and the bag are detected, timing is started, if the time of no person exceeds 30 minutes, the seat is judged to be occupied maliciously, and a green light is turned on.
In an embodiment of the present invention, in step S1, the training of YOLO V5 includes the following steps;
step A1, firstly, constructing a data set containing a table, a chair and students, and then amplifying the data set by affine transformation and rotation to generate a data set which is large enough and has high picture quality;
a2, building a neural network model required by a YOLO V5 framework, adopting a pitorch network framework and a Relu activation function as an activation function, and solving the problem of neuron death when an input value is a negative value;
step A3, dividing data into training sets: and (4) verification set: test set = 8: 1: adding an Auto Learning Bounding Box-adaptive anchor frame, analyzing a user-defined data set by adopting a k-means and a genetic Learning algorithm, and obtaining a preset anchor frame suitable for predicting an object boundary frame in the user-defined data set, wherein the frame is automatically learned based on training data.
In an embodiment of the invention, by using Darknet-53 as a backbone network, 3 feature layers with different scales are respectively 13 × 13, 26 × 26 and 52 × 52, and YOLO V5 firstly sets 3 prior frames for each downsampling scale so as to cluster to obtain prior frames with 9 sizes; in the entire YOLO V5 structure, there are no pooling layers and full connectivity layers, and downsampling of the network is achieved by setting the stride of the convolution to 3; a Deepsort algorithm is fused on the basis of target identification based on the current-stage YOLO V5 algorithm, and prediction heads are integrated and applied to unmanned aerial vehicle shooting, so that the technology that the target can be accurately positioned and continuously tracked in a high-density scene is finally achieved.
Compared with the prior art, the invention has the following beneficial effects: the method comprises the steps of monitoring whether a seat is occupied on a table in a library or not by arranging a monitoring camera in the library and utilizing a YOLO V5 model, sending a video frame acquired by the camera into the YOLO V5 model, fusing a Deepsort algorithm on the basis of identifying a target based on the current YOLO V5 algorithm, integrating prediction heads, applying the detection heads to unmanned aerial vehicle shooting, and finally achieving the technology of accurately positioning the target in a high-density scene and continuously tracking so as to achieve the purpose of real-time monitoring.
Drawings
FIG. 1 is a schematic diagram of the network structure of YOLO V5;
FIG. 2 is a formula for a loss function of the YOLO V5 model;
FIG. 3 is a schematic workflow diagram of an embodiment of the present invention;
FIG. 4 illustrates the identification of whether a chair is empty in a restaurant according to the present invention;
fig. 5 shows the result of recognition of whether or not an occupied seat article is present when an empty seat is detected.
Detailed Description
The technical scheme of the invention is specifically explained below with reference to the accompanying drawings.
The invention relates to a campus library occupation detection method based on YOLO v5, which adopts a K-means fused YOLO 5 detection method to obtain a more reasonable anchor frame algorithm and improve the accuracy of classified target detection, and the method is concretely implemented by the following steps:
step S1, obtaining scene images from the monitoring videos of the monitoring areas of the cameras in the library, and detecting tables and chairs in the images according to a deep learning target detection algorithm framework YOLO 5;
s2, obtaining more reasonable 9 anchors frames needed by YOLO V5 by using a K-means clustering algorithm;
step S3, after the chair and the person are detected, the IOU ratio of the coincidence degree between the images of the chair and the person is calculated to judge whether the chair is occupied or not;
step S4, judging whether the distance is smaller than a threshold value, if so, judging that a person is on the seat, lighting a red light, and if not, continuing to detect the book and the bag;
step S5, if no book or bag is detected, then it is determined that no person is present and the position is sitting, and a green light is turned on;
and step S6, if the book and the bag are detected, timing is started, if the time of no person exceeds 30 minutes, the seat is judged to be occupied maliciously, and a green light is turned on.
The following are specific embodiments of the present invention.
The campus library seat occupation detection method based on the YOLO V5 is characterized in that an end-to-end deep learning target detection algorithm YOLO V5 is used for detection of students, tables, chairs, books and bags, a model is optimized by adopting a K-means algorithm, and detection performance is improved. The YoLO V5 network model is shown in FIG. 1.
Specifically, the YOLO V5 model adopts ReLU as an activation function and is trained by an end-to-end method, and the YOLO V5 model adopts a loss function when performing a gradient descent method as shown in fig. 2:
the first part and the second part are responsible for predicting bbox (boundary box) of an object, the first part represents error values of center point coordinates and ground truth center point mislabels obtained by forward propagation of the image along the neural network, and the second part measures error values of frame width height and ground truth width height obtained by forward propagation of the image along the neural network; the third part represents an error value of the confidence coefficient of the prediction frame containing the target object, and the confidence coefficient of the prediction frame containing the target object reaches 1 after training; the fourth part represents an error value of the confidence coefficient of the prediction frame without the target object, and the confidence coefficient of the prediction widening without the target object is trained to reach 0; the fifth part is a mesh prediction classification error term that contains the target object.
The specific implementation steps will be described with reference to the workflow diagram of fig. 3:
step 1, obtaining a scene image from a monitoring video of a monitoring area by a monitoring camera;
step 3, detecting students and chairs in the image according to a deep learning target detection algorithm framework YOLO 5, and identifying and outputting by using a rectangular frame;
step 4, judging whether a person sits on the chair or not according to the IOU ratio of the chair to the person image prior frame;
step 5, if the chair is judged to be occupied, controlling a red light to be on, and if the chair is judged to be unoccupied, detecting books and bags;
step 6, if the person is detected, timing for 30 minutes, and if the person is not detected in more than 30 minutes, judging that the behavior is seat occupation;
fig. 4 is a diagram of detecting whether a person is present in a seat: detecting the IOU ratio map of the person, the chair and the prior frames of the person and the chair through different detection frames, and judging whether the person sits on the chair or not according to the ratio;
fig. 5 is a diagram showing the effect of detecting whether there is an object occupying the seat on the table when no person is present.
The above are preferred embodiments of the present invention, and all changes made according to the technical scheme of the present invention that produce functional effects do not exceed the scope of the technical scheme of the present invention belong to the protection scope of the present invention.
Claims (3)
1. A campus library occupation detection method based on YOLO v5 is characterized in that a K-means fused YOLO 5 detection method is adopted to obtain a more reasonable anchor frame algorithm and improve the accuracy of classification target detection, and the method is concretely implemented by the following steps:
step S1, obtaining scene images from the monitoring videos of the monitoring areas of the cameras in the library, and detecting tables and chairs in the images according to a deep learning target detection algorithm framework YOLO 5;
s2, obtaining more reasonable 9 anchors frames needed by YOLO V5 by using a K-means clustering algorithm;
step S3, after the chair and the person are detected, the IOU ratio of the coincidence degree between the images of the chair and the person is calculated to judge whether the chair is occupied or not;
step S4, judging whether the distance is smaller than a threshold value, if so, judging that a person is on the seat, lighting a red light, and if not, continuing to detect the book and the bag;
step S5, if no book or bag is detected, then it is determined that no person is present and the position is sitting, and a green light is turned on;
and step S6, if the book and the bag are detected, timing is started, if the time of no person exceeds 30 minutes, the seat is judged to be occupied maliciously, and a green light is turned on.
2. The method for detecting occupancy in campus library based on YOLO V5 as claimed in claim 1, wherein in step S1, the training of YOLO V5 includes the following steps;
step A1, firstly, constructing a data set containing a table, a chair and students, and then amplifying the data set by affine transformation and rotation to generate a data set which is large enough and has high picture quality;
a2, building a neural network model required by a YOLO V5 framework, adopting a pitorch network framework and a Relu activation function as an activation function, and solving the problem of neuron death when an input value is a negative value;
step A3, dividing data into training sets: and (4) verification set: test set = 8: 1: adding an Auto Learning Bounding Box-adaptive anchor frame, analyzing a user-defined data set by adopting a k-means and a genetic Learning algorithm, and obtaining a preset anchor frame suitable for predicting an object boundary frame in the user-defined data set, wherein the frame is automatically learned based on training data.
3. The method for detecting occupancy of a campus library based on YOLO V5 as claimed in claim 2, wherein Darknet-53 is used as backbone network, the feature layers of 3 different scales are respectively 13 × 13, 26 × 26 and 52 × 52, YOLO V5 firstly sets 3 prior frames for each downsampling scale so as to cluster to obtain 9 prior frames; in the entire YOLO V5 structure, there are no pooling layers and full connectivity layers, and downsampling of the network is achieved by setting the stride of the convolution to 3; a Deepsort algorithm is fused on the basis of target identification based on the current-stage YOLO V5 algorithm, and prediction heads are integrated and applied to unmanned aerial vehicle shooting, so that the technology that the target can be accurately positioned and continuously tracked in a high-density scene is finally achieved.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111545932.0A CN114170573A (en) | 2021-12-17 | 2021-12-17 | School library seat occupation detection method based on YOLO v5 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111545932.0A CN114170573A (en) | 2021-12-17 | 2021-12-17 | School library seat occupation detection method based on YOLO v5 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114170573A true CN114170573A (en) | 2022-03-11 |
Family
ID=80487020
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111545932.0A Pending CN114170573A (en) | 2021-12-17 | 2021-12-17 | School library seat occupation detection method based on YOLO v5 |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114170573A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117826692A (en) * | 2024-03-04 | 2024-04-05 | 深圳市昊洋智能有限公司 | Intelligent classroom multimedia management platform utilizing Internet of things for centralized management and control |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109710567A (en) * | 2018-12-29 | 2019-05-03 | 滨州学院 | A kind of library's room for individual study is anti-to account for an information processing system and method |
| CN111127727A (en) * | 2020-02-20 | 2020-05-08 | 山东科技大学 | Library's intelligence is prevented occupying seat and cleaning device |
| CN111598173A (en) * | 2020-05-19 | 2020-08-28 | 腾讯科技（深圳）有限公司 | Seat occupation preventing method and device |
| CN112488006A (en) * | 2020-12-05 | 2021-03-12 | 东南大学 | Target detection algorithm based on wheat image |
| CN112883978A (en) * | 2021-03-08 | 2021-06-01 | 宁波工程学院 | Seat identification method based on neural network |
-
2021
- 2021-12-17 CN CN202111545932.0A patent/CN114170573A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109710567A (en) * | 2018-12-29 | 2019-05-03 | 滨州学院 | A kind of library's room for individual study is anti-to account for an information processing system and method |
| CN111127727A (en) * | 2020-02-20 | 2020-05-08 | 山东科技大学 | Library's intelligence is prevented occupying seat and cleaning device |
| CN111598173A (en) * | 2020-05-19 | 2020-08-28 | 腾讯科技（深圳）有限公司 | Seat occupation preventing method and device |
| CN112488006A (en) * | 2020-12-05 | 2021-03-12 | 东南大学 | Target detection algorithm based on wheat image |
| CN112883978A (en) * | 2021-03-08 | 2021-06-01 | 宁波工程学院 | Seat identification method based on neural network |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117826692A (en) * | 2024-03-04 | 2024-04-05 | 深圳市昊洋智能有限公司 | Intelligent classroom multimedia management platform utilizing Internet of things for centralized management and control |
| CN117826692B (en) * | 2024-03-04 | 2024-05-24 | 深圳市昊洋智能有限公司 | Intelligent classroom multimedia management platform utilizing Internet of things for centralized management and control |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108388888B (en) | Vehicle identification method and device and storage medium | |
| CN105574550B (en) | A kind of vehicle identification method and device | |
| CN109284733B (en) | Shopping guide negative behavior monitoring method based on yolo and multitask convolutional neural network | |
| JP2019075116A (en) | Method for acquiring bounding box corresponding to object on image by using cnn (convolutional neural network) including tracking network | |
| CN108921083A (en) | Illegal flowing street pedlar recognition methods based on deep learning target detection | |
| CN108416250A (en) | Demographic method and device | |
| WO2021047306A1 (en) | Abnormal behavior determination method and apparatus, terminal, and readable storage medium | |
| CN108564673A (en) | A kind of check class attendance method and system based on Global Face identification | |
| CN109359696A (en) | A kind of vehicle money recognition methods, system and storage medium | |
| CN106570497A (en) | Text detection method and device for scene image | |
| CN107229929A (en) | A kind of license plate locating method based on R CNN | |
| US20190236738A1 (en) | System and method for detection of identity fraud | |
| CN109255375A (en) | Panoramic picture method for checking object based on deep learning | |
| CN104134364B (en) | Real-time traffic sign identification method and system with self-learning capacity | |
| CN108875754B (en) | Vehicle re-identification method based on multi-depth feature fusion network | |
| CN109858375A (en) | Living body faces detection method, terminal and computer readable storage medium | |
| CN104376303B (en) | A kind of vehicle imaging method in the case of low resolution | |
| CN108764096A (en) | A kind of pedestrian weight identifying system and method | |
| CN109948616A (en) | Image detecting method, device, electronic equipment and computer readable storage medium | |
| CN110827432B (en) | Class attendance checking method and system based on face recognition | |
| CN109977876A (en) | Image-recognizing method, calculates equipment, system and storage medium at device | |
| CN107784273B (en) | Classroom seat distribution prediction method based on student social relationship model | |
| CN109961013A (en) | Recognition methods, device, equipment and the computer readable storage medium of lane line | |
| CN106886778A (en) | A kind of car plate segmentation of the characters and their identification method under monitoring scene | |
| CN110490214A (en) | The recognition methods and system of image, storage medium and processor |
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 |