CN112633249A - Embedded pedestrian flow detection method based on light deep learning framework - Google Patents
Embedded pedestrian flow detection method based on light deep learning framework Download PDFInfo
- Publication number
- CN112633249A CN112633249A CN202110008192.0A CN202110008192A CN112633249A CN 112633249 A CN112633249 A CN 112633249A CN 202110008192 A CN202110008192 A CN 202110008192A CN 112633249 A CN112633249 A CN 112633249A
- Authority
- CN
- China
- Prior art keywords
- deep learning
- embedded
- light
- method based
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses an embedded pedestrian flow detection method based on a light deep learning framework, which comprises the following steps of: s1, building a hardware platform; s2, sampling, labeling and standardizing pedestrians by using a hardware platform, collecting training data and preprocessing the data; s3, improving a light SSD300 deep learning framework, and building a people flow detection deep learning model; s4, training the built people flow detection deep learning model according to the collected data and transplanting the model to a hardware platform; and S5, analyzing the result and visually displaying the result. According to the invention, embedded equipment is selected to connect a camera as a detection platform, a light improved SSD300 people flow detection deep learning model is carried, people flow detection is carried out on the collected images, and people flow data are transmitted to a cloud server in real time for a terminal to display in real time and analyze results. The invention has the advantages of simple and convenient hardware installation and maintenance, low cost and high detection result precision.
Description
Technical Field
The invention relates to the technical field of intelligent monitoring, in particular to an embedded pedestrian flow detection method based on a light deep learning framework.
Background
People flow detection is a research hotspot in the field of intelligent monitoring and is also an objective requirement for building smart cities and digital scenic spots. Through the real-time accurate estimation of crowd density and change thereof, can in time carry out effectual reposition of redundant personnel, mediation and control to the crowd to improve road utilization and visitor's efficiency of visiting, prevent that a large amount of visitors from blocking up in the scenic spot in the condition of individual sight spot, the accident is trampled in the prevention, avoids unnecessary time and loss of lives and property. In addition, the real-time accurate monitoring of the flow of people is realized, and the method has important significance for commercial information acquisition, public security prevention and control, reasonable social resource allocation and the like.
At present, the commonly adopted people flow monitoring and counting methods mainly include gate counting, infrared detection, machine vision and other technologies, and the methods can be divided into two main types, namely contact methods and non-contact methods. The contact method needs to arrange a mechanical gate or a pressure-sensitive pedal at a monitoring point, and has the limitations of higher equipment installation and operation and maintenance cost or easiness in damage and the like; in the non-contact method, the accuracy of statistics on conditions such as multiple people and shielding is not high by an infrared-based detection statistical technology, people flow monitoring based on machine vision becomes a hot research branch, and the method has the advantages that an original image is easy to obtain, numerous machine learning algorithms can be used for reference, and the like, but the energy consumption is high, and the cost advantage is not obvious. In view of the above problems of people flow rate detection, it is urgently needed to provide an embedded people flow rate detection method based on a light deep learning framework, so as to solve the deficiencies in the prior art.
Disclosure of Invention
The invention aims to provide an embedded pedestrian flow detection method based on a light deep learning framework, which is used for solving the problems of high cost, high energy consumption and the like of the conventional pedestrian flow detection.
An embedded pedestrian flow detection method based on a light deep learning framework comprises the following steps:
s1, selecting embedded equipment and a camera as a detection platform, connecting the detection platform with a cloud server through a wireless network, and building a system hardware platform;
s2, under a specific scene, sampling, labeling and standardizing pedestrians through the hardware platform, collecting training data and preprocessing the data;
s3, improving a light SSD300 deep learning framework, and building a people flow detection deep learning model;
s4, training the built people flow detection deep learning model according to the collected training data, transplanting the training data to the embedded equipment, and reasoning people flow in real time;
s5, building a web cloud service, and transmitting the people flow information acquired by the embedded equipment to the web cloud service by using a wireless network as a data transmission carrier, wherein the web cloud service is connected with each visual terminal in real time and carries out result analysis and display.
Optionally, in step S2, the application of the hardware platform to sample, label, and standardize the pedestrian, collect training data, and preprocess the data includes:
s201, performing video sampling by using a camera, and independently storing each frame of image contained in a video as an original image;
s202, preprocessing the acquired image data, and modifying the size of the image to 640x 480;
s203, for each image, marking each person by using a rectangular frame, requiring to completely wrap the pedestrian, recording and storing coordinates of the upper left and the lower right of the rectangular frame by using a JSON file, wherein the data format is VOC.
Optionally, in step S3, the light SSD300 deep learning framework is improved, and a people flow rate detection deep learning model is built, which specifically includes:
s301, changing a volume base layer in vgg16 as a basic network into a depth separable volume layer;
s302, adding a layer of cavity convolution after the seventh layer in the basic layer;
s303, changing the multi-scale feature maps in the SSD330 into four, respectively detecting size targets, wherein the sizes of the feature maps are respectively 80x60, 40x30, 20x15 and 10x8 in the input picture resolution of 640x 480;
s304, the output is divided into two parts, wherein one part is socre used for predicting classification confidence, the scale is 17640x2, and the part is the confidence of the background and the pedestrian; the other part is boxes with the scale of 17640x4, and is the center point, width and height of the prediction box.
Optionally, in step S4, training the constructed human traffic detection model according to the collected data, and transplanting the human traffic detection model to a hardware platform, specifically including:
s401, training the built people flow detection deep learning model according to the collected data;
s402, transplanting the trained people flow detection deep learning model to the hardware platform;
and S403, reasoning the human flow in real time.
Optionally, in step S403, the real-time reasoning about the human traffic specifically includes:
step 1, reading a frame of picture in a camera video stream;
and 4, filtering the detection result by using an NMS algorithm, wherein the IOU threshold value is 0.3.
Optionally, in step S5, the analyzing result is visually displayed, and the method specifically includes:
s501, uploading the acquired people flow reasoning result to the cloud server in real time through the wireless network;
s502, developing a web application program by using a flash framework, and deploying the web application program to a cloud server by using Gunicorn + Nginx;
and S503, drawing a current pedestrian volume line graph, dynamically reflecting the crowding degree of the pedestrians through different colors of the progress bar according to the current pedestrian volume quantity, and sending the crowding degree to the terminal for displaying.
Optionally, the hardware platform in step S1 includes an ARM architecture embedded development board raspberry pi 3B +, a camera, and a cloud server, where the ARM architecture embedded development board raspberry pi 3B + is electrically connected to the camera, the ARM architecture embedded development board raspberry pi 3B + has a built-in wireless module, and the wireless module is used for communicating with the cloud server.
Optionally, the model of the camera is OV 5647.
According to the specific steps provided by the invention, the invention discloses the following technical effects: according to the embedded pedestrian volume detection method based on the light deep learning framework, an embedded device is selected to be connected with a camera as a detection platform, a light improved SSD300 pedestrian volume detection deep learning model is carried, pedestrian volume detection is carried out on collected images, pedestrian volume data are transmitted to a cloud server in real time, and the terminal displays the pedestrian volume data in real time and analyzes results. The invention has the advantages of simple and convenient hardware installation and maintenance, low cost and high detection result precision.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flow chart of an embedded pedestrian flow detection method based on a light deep learning framework according to the invention;
FIG. 2 is a structural diagram of a hardware platform of the embedded human flow detection method based on a light deep learning framework;
FIG. 3 is a preprocessed image annotation data graph of the embedded pedestrian volume detection method based on the light deep learning framework;
FIG. 4 is a schematic diagram of an improved SSD300 model structure of the embedded people flow detection method based on the light deep learning framework of the present invention;
FIG. 5 is a schematic diagram of a pedestrian volume detection result of an SSD300 model structure of the embedded pedestrian volume detection method based on the light deep learning framework of the present invention;
fig. 6 is a diagram of an example of functions provided by a web cloud service of the embedded people flow detection method based on the light deep learning framework.
Detailed Description
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 only a part of the embodiments of the present invention, and not all of the embodiments. 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.
The invention aims to provide an embedded pedestrian flow detection method based on a light deep learning framework, which is used for solving the problems of high cost, high energy consumption and the like of the conventional pedestrian flow detection.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
The invention provides an embedded pedestrian flow detection method based on a light deep learning framework, which comprises the following specific steps of:
s1, selecting embedded equipment and a camera as a detection platform, connecting the detection platform with a cloud server through a wireless network, and building a system hardware platform;
as shown in fig. 2, the hardware platform includes an ARM architecture embedded development board raspberry pi 3B +, a camera and a cloud server, the ARM architecture embedded development board raspberry pi 3B + is electrically connected to the camera, the ARM architecture embedded development board raspberry pi 3B + has a built-in wireless module, the wireless module is used for communicating with the cloud server, and the model of the camera is OV 5647;
s2, in a specific scenario, sampling, labeling, and standardizing the pedestrian through the hardware platform, collecting training data, and preprocessing the data, as shown in fig. 3, specifically including:
s201, performing video sampling by using a camera, and independently storing each frame of image contained in a video as an original image;
s202, preprocessing the acquired image data, and modifying the size of the image to 640x 480;
s203, labeling each person by using a rectangular frame aiming at each image, requiring to completely wrap the pedestrian, recording and storing coordinates of the upper left and the lower right of the rectangular frame by using a JSON file, wherein the data format is VOC;
s3, improving a light SSD300 deep learning framework, and building a people flow detection deep learning model, as shown in FIG. 4, specifically comprising:
s301, changing a volume base layer in vgg16 as a basic network into a depth separable volume layer, and greatly reducing convolution kernel parameters under the condition of obtaining unchanged feature diagram size and channel number;
s302, adding a layer of cavity convolution after the seventh layer in the basic layer;
s303, changing the multi-scale feature maps in the SSD300 into four, respectively detecting size targets, wherein the sizes of the feature maps are respectively 80x60, 40x30, 20x15 and 10x8 in the input picture resolution of 640x 480;
s304, the output is divided into two parts, wherein one part is socre used for predicting classification confidence, the scale is 17640x2, and the part is the confidence of the background and the pedestrian; the other part is boxes with the scale of 17640x4, and is the central point, width and height of the prediction box; in the improved people flow detection deep learning model of the SSD300, a loss function is divided into two parts, one part is an error of score, and cross entropy calculation is used; the other part is the error of the boxes, and is calculated by using Smooth L1 Loss;
s4, training the built human traffic detection deep learning model according to the collected training data and transplanting the human traffic detection deep learning model to the hardware platform, wherein the training specifically comprises the following steps:
s401, training the built people flow detection deep learning model according to the collected data;
s402, transplanting the trained people flow detection deep learning model to the hardware platform;
s403, reasoning the human flow in real time;
in step S403, the real-time reasoning of the human traffic specifically includes:
step 1, reading a frame of picture in a camera video stream;
s5, building a web cloud service, and transmitting the people flow information obtained by the embedded device to the web cloud service by using a wireless network as a data transmission carrier, where the web cloud service is connected with each of the visual terminals in real time, and performs result analysis and display, as shown in fig. 6, the method specifically includes:
s501, uploading the acquired people flow reasoning result to a cloud server in real time through the wireless module;
s502, developing a web application program by using a flash framework, and deploying the web application program to a cloud server by using Gunicorn + Nginx;
s503, drawing a current pedestrian volume line graph, dynamically reflecting the crowding degree of the pedestrians through different colors of the progress bar according to the current pedestrian volume quantity, and sending the crowding degree to a terminal for displaying;
the web cloud service function in step S502 specifically includes: monitoring an access request of the embedded equipment, triggering a popup window after monitoring the access request, and displaying an ip and a port number of the embedded equipment; multithreading design, receiving people flow data from the embedded equipment in real time; refreshing in real time by using a millimeter-scale timer, drawing a current pedestrian volume line graph and sending the line graph to a terminal for displaying; the crowding degree of the pedestrian is dynamically reflected by different colors (green represents rare, blue represents normal and red represents crowding) of the progress bar according to the current pedestrian flow quantity.
According to the embedded pedestrian volume detection method based on the light deep learning framework, an embedded device is selected to be connected with a camera as a detection platform, a light improved SSD300 pedestrian volume detection deep learning model is carried, pedestrian volume detection is carried out on collected images, pedestrian volume data are transmitted to a cloud server in real time, and the terminal displays the pedestrian volume data in real time and analyzes results. The invention has the advantages of simple and convenient hardware installation and maintenance, low cost and high detection result precision.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (8)
1. An embedded pedestrian flow detection method based on a light deep learning framework is characterized by comprising the following steps:
s1, selecting embedded equipment and a camera as a detection platform, connecting the detection platform with a cloud server through a wireless network, and building a system hardware platform;
s2, under a specific scene, sampling, labeling and standardizing pedestrians through the hardware platform, collecting training data and preprocessing the data;
s3, improving a light SSD300 deep learning framework, and building a people flow detection deep learning model;
s4, training the built people flow detection deep learning model according to the collected training data, transplanting the training data to the embedded equipment, and reasoning people flow in real time;
s5, building a web cloud service, and transmitting the people flow information acquired by the embedded equipment to the web cloud service by using a wireless network as a data transmission carrier, wherein the web cloud service is connected with each visual terminal in real time and carries out result analysis and display.
2. The embedded pedestrian flow detection method based on the light deep learning framework of claim 1, wherein in the step S2, the steps of sampling, labeling and standardizing pedestrians through a hardware platform, collecting training data and preprocessing the data specifically include:
s201, performing video sampling by using a camera, and independently storing each frame of image contained in a video as an original image;
s202, preprocessing the acquired image data, and modifying the size of the image to 640x 480;
s203, for each image, marking each person by using a rectangular frame, requiring to completely wrap the pedestrian, recording and storing coordinates of the upper left and the lower right of the rectangular frame by using a JSON file, wherein the data format is VOC.
3. The embedded pedestrian volume detection method based on the light-weight deep learning framework as claimed in claim 1, wherein in step S3, the light-weight SSD300 deep learning framework is improved, and a pedestrian volume detection deep learning model is built, specifically including:
s301, changing a volume base layer in vgg16 as a basic network into a depth separable volume layer;
s302, adding a layer of cavity convolution after the seventh layer in the basic layer;
s303, changing the multi-scale feature maps in the SSD300 into four, respectively detecting size targets, wherein the sizes of the feature maps are respectively 80x60, 40x30, 20x15 and 10x8 in the input picture resolution of 640x 480;
s304, the output is divided into two parts, wherein one part is socre used for predicting classification confidence, the scale is 17640x2, and the part is the confidence of the background and the pedestrian; the other part is boxes with the scale of 17640x4, and is the center point, width and height of the prediction box.
4. The embedded human traffic detection method based on the light deep learning framework according to claim 1, wherein in the step S4, the built human traffic detection model is trained according to the collected data and transplanted to a hardware platform, and specifically comprises:
s401, training the built people flow detection deep learning model according to the collected data;
s402, transplanting the trained people flow detection deep learning model to the hardware platform;
and S403, reasoning the human flow in real time.
5. The embedded human traffic detection method based on the light deep learning framework according to claim 4, wherein in the step S403, the real-time inference of human traffic specifically comprises:
step 1, reading a frame of picture in a camera video stream;
step 2, standardizing the picture, namely subtracting a mean value 127 from each pixel in the picture, dividing the mean value by a normal value 128, and performing format conversion on the picture, namely converting the channel arrangement BGR of the picture into RGB;
step 3, establishing a light deep learning model pusher, and transmitting the picture into the pusher for reasoning to obtain a detection result;
and 4, filtering the detection result by using an NMS algorithm, wherein the IOU threshold value is 0.3.
6. The embedded human traffic detection method based on the light deep learning framework as claimed in claim 1, wherein the step S5 analyzes the result and performs visual display, specifically comprising:
s501, uploading the acquired people flow reasoning result to the cloud server in real time through the wireless network;
s502, developing a web application program by using a flash framework, and deploying the web application program to a cloud server by using Gunicorn + Nginx;
and S503, drawing a current pedestrian volume line graph, dynamically reflecting the crowding degree of the pedestrians through different colors of the progress bar according to the current pedestrian volume quantity, and sending the crowding degree to the terminal for displaying.
7. The embedded pedestrian volume detection method based on the light deep learning framework is characterized in that the hardware platform in the step S1 includes an ARM architecture embedded development board raspberry pi 3B +, a camera and a cloud server, the ARM architecture embedded development board raspberry pi 3B + is electrically connected to the camera, the ARM architecture embedded development board raspberry pi 3B + has a built-in wireless module, and the wireless module is used for communicating with the cloud server.
8. The embedded pedestrian flow detection method based on the light deep learning framework is characterized in that the model of the camera is OV 5647.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110008192.0A CN112633249A (en) | 2021-01-05 | 2021-01-05 | Embedded pedestrian flow detection method based on light deep learning framework |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110008192.0A CN112633249A (en) | 2021-01-05 | 2021-01-05 | Embedded pedestrian flow detection method based on light deep learning framework |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112633249A true CN112633249A (en) | 2021-04-09 |
Family
ID=75290700
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110008192.0A Pending CN112633249A (en) | 2021-01-05 | 2021-01-05 | Embedded pedestrian flow detection method based on light deep learning framework |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112633249A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113780053A (en) * | 2021-07-05 | 2021-12-10 | 青海大学 | People stream density monitoring system based on development board |
| CN114419473A (en) * | 2021-11-17 | 2022-04-29 | 中国电子科技集团公司第三十八研究所 | Deep learning real-time target detection method based on embedded equipment |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160307420A1 (en) * | 2013-03-15 | 2016-10-20 | Bruno Delean | Vision based system for detecting a breach of security in a monitored location |
| CN107204050A (en) * | 2017-05-27 | 2017-09-26 | 西安电子科技大学 | A kind of WIFI sent based on raspberry is registered system and method |
| CN108229452A (en) * | 2018-03-20 | 2018-06-29 | 东北大学 | People counting device and method based on deep neural network and dsp chip |
| CN109614941A (en) * | 2018-12-14 | 2019-04-12 | 中山大学 | A kind of embedded crowd density estimation method based on convolutional neural networks model |
| CN109815882A (en) * | 2019-01-21 | 2019-05-28 | 南京行者易智能交通科技有限公司 | A kind of subway carriage intensity of passenger flow monitoring system and method based on image recognition |
| CN110942055A (en) * | 2019-12-31 | 2020-03-31 | 北京市商汤科技开发有限公司 | State identification method, device and equipment for display area and storage medium |
| CN111199220A (en) * | 2020-01-21 | 2020-05-26 | 北方民族大学 | Lightweight deep neural network method for people detection and people counting in elevator |
| CN111629336A (en) * | 2020-06-05 | 2020-09-04 | 腾讯科技（深圳）有限公司 | Method, device and equipment for determining number of people in target area and storage medium |
| CN111640135A (en) * | 2020-05-25 | 2020-09-08 | 台州智必安科技有限责任公司 | TOF camera pedestrian counting method based on hardware front end |
| CN112270381A (en) * | 2020-11-16 | 2021-01-26 | 电子科技大学 | People flow detection method based on deep learning |
-
2021
- 2021-01-05 CN CN202110008192.0A patent/CN112633249A/en active Pending
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160307420A1 (en) * | 2013-03-15 | 2016-10-20 | Bruno Delean | Vision based system for detecting a breach of security in a monitored location |
| CN107204050A (en) * | 2017-05-27 | 2017-09-26 | 西安电子科技大学 | A kind of WIFI sent based on raspberry is registered system and method |
| CN108229452A (en) * | 2018-03-20 | 2018-06-29 | 东北大学 | People counting device and method based on deep neural network and dsp chip |
| CN109614941A (en) * | 2018-12-14 | 2019-04-12 | 中山大学 | A kind of embedded crowd density estimation method based on convolutional neural networks model |
| CN109815882A (en) * | 2019-01-21 | 2019-05-28 | 南京行者易智能交通科技有限公司 | A kind of subway carriage intensity of passenger flow monitoring system and method based on image recognition |
| CN110942055A (en) * | 2019-12-31 | 2020-03-31 | 北京市商汤科技开发有限公司 | State identification method, device and equipment for display area and storage medium |
| CN111199220A (en) * | 2020-01-21 | 2020-05-26 | 北方民族大学 | Lightweight deep neural network method for people detection and people counting in elevator |
| CN111640135A (en) * | 2020-05-25 | 2020-09-08 | 台州智必安科技有限责任公司 | TOF camera pedestrian counting method based on hardware front end |
| CN111629336A (en) * | 2020-06-05 | 2020-09-04 | 腾讯科技（深圳）有限公司 | Method, device and equipment for determining number of people in target area and storage medium |
| CN112270381A (en) * | 2020-11-16 | 2021-01-26 | 电子科技大学 | People flow detection method based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| ANDRES HEREDIA 等: "Video Processing inside embedded devices using SSD-Mobilenet to count mobility actors", 《2019 IEEE COLOMBIAN CONFERENCE ON APPLICATIONS IN COMPUTATIONAL INTELLIGENCE》 * |
| WEN ZHENG ZHAO 等: "Rapid and Accurate Object Detection on Drone based Embedded Devices with Dilated, Deformableand Pyramid Convolution", 《PROC.SPIE 11427, SECOND TARGET RECOGNITION AND ARTIFICIAL INTELLIGENCE SUMMIT FORUM》 * |
| 郑国书 等: "基于深度学习SSD模型的视频室内人数统计", 《工业控制计算机》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113780053A (en) * | 2021-07-05 | 2021-12-10 | 青海大学 | People stream density monitoring system based on development board |
| CN114419473A (en) * | 2021-11-17 | 2022-04-29 | 中国电子科技集团公司第三十八研究所 | Deep learning real-time target detection method based on embedded equipment |
| CN114419473B (en) * | 2021-11-17 | 2024-04-16 | 中国电子科技集团公司第三十八研究所 | Deep learning real-time target detection method based on embedded equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110428522B (en) | Intelligent security system of wisdom new town | |
| WO2022126669A1 (en) | Subway pedestrian flow network fusion method based on video pedestrian recognition, and pedestrian flow prediction method | |
| CN116824859B (en) | Intelligent traffic big data analysis system based on Internet of things | |
| KR102122859B1 (en) | Method for tracking multi target in traffic image-monitoring-system | |
| CN104766086B (en) | The monitoring and managing method and system of a kind of way mark | |
| CN108965826A (en) | Monitoring method, device, processing equipment and storage medium | |
| CN110555420B (en) | Fusion model network and method based on pedestrian regional feature extraction and re-identification | |
| CN106127292B (en) | Flow method of counting and equipment | |
| CN113096158A (en) | Moving object identification method and device, electronic equipment and readable storage medium | |
| CN112633249A (en) | Embedded pedestrian flow detection method based on light deep learning framework | |
| CN112668375B (en) | Tourist distribution analysis system and method in scenic spot | |
| CN104134067A (en) | Road vehicle monitoring system based on intelligent visual Internet of Things | |
| CN112449147B (en) | Video cluster monitoring system of photovoltaic power station and image processing method thereof | |
| CN113378668A (en) | Method, device and equipment for determining accumulated water category and storage medium | |
| CN112712707A (en) | Vehicle carbon emission monitoring system and method | |
| CN115311591A (en) | Early warning method and device for abnormal behaviors and intelligent camera | |
| CN111428653B (en) | Pedestrian congestion state judging method, device, server and storage medium | |
| US20210104059A1 (en) | Method for size estimation by image recognition of specific target using given scale | |
| CN111950491A (en) | Personnel density monitoring method and device and computer readable storage medium | |
| CN109740527B (en) | Image processing method in video frame | |
| CN114222163A (en) | Intelligent video analysis system for business hall | |
| CN117273405B (en) | Method for managing scenic spot by using array computing vision | |
| CN113239718A (en) | Campus people flow rate statistical method | |
| CN117475269A (en) | Abnormal signal identification method and system based on video monitoring system equipment | |
| CN117475352A (en) | River course parabolic behavior detection method based on deep learning |
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 |