CN113537007A - Non-worker intrusion detection and alarm method and device applied to railway platform - Google Patents
Non-worker intrusion detection and alarm method and device applied to railway platform Download PDFInfo
- Publication number
- CN113537007A CN113537007A CN202110751325.3A CN202110751325A CN113537007A CN 113537007 A CN113537007 A CN 113537007A CN 202110751325 A CN202110751325 A CN 202110751325A CN 113537007 A CN113537007 A CN 113537007A
- Authority
- CN
- China
- Prior art keywords
- alarm
- video
- intrusion detection
- target
- detection model
- 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 100
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 9
- 230000001133 acceleration Effects 0.000 claims description 8
- 230000006835 compression Effects 0.000 claims description 8
- 238000007906 compression Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 8
- 238000013526 transfer learning Methods 0.000 claims description 5
- 230000001960 triggered effect Effects 0.000 claims description 5
- 238000013138 pruning Methods 0.000 claims description 4
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 230000005012 migration Effects 0.000 claims description 3
- 230000001502 supplementing effect Effects 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 4
- 238000007689 inspection Methods 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 description 12
- 238000007726 management method Methods 0.000 description 9
- 238000012544 monitoring process Methods 0.000 description 8
- 238000004891 communication Methods 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 5
- 238000013145 classification model Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000013461 design Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 230000007123 defense Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 230000004304 visual acuity Effects 0.000 description 1
- 238000012800 visualization 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
- 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
- G06N3/045—Combinations of networks
-
- 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
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/292—Multi-camera tracking
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
- G08B13/19602—Image analysis to detect motion of the intruder, e.g. by frame subtraction
- G08B13/19608—Tracking movement of a target, e.g. by detecting an object predefined as a target, using target direction and or velocity to predict its new position
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30232—Surveillance
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- Alarm Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a non-worker intrusion detection and alarm method and device applied to a railway platform. The method comprises the steps of obtaining a real-time video of a station end area of a station platform; detecting and tracking a target in the video by using an intrusion detection model, and judging whether to trigger an alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff feature detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm. According to the method and the device provided by the invention, through using the intrusion detection model, workers and non-workers can be distinguished, real-time online analysis of dozens of paths of high-definition videos operated by one image processor can be realized, false alarm caused by regular inspection of the workers is solved to a certain extent, meanwhile, the pressure of safety management of the passenger station when passenger flow increases is effectively relieved, and the system has strong reliability and low cost.
Description
Technical Field
The invention relates to the field of operation safety monitoring and prevention and control of railway passenger stations, in particular to a non-worker intrusion detection and alarm method and device applied to a railway platform.
Background
The development of railway operation lines, particularly high-speed railways, is receiving more and more attention, and is favored by people with more travel demands. The railway passenger station is used as a primary interactive window for the traveling of high-speed railways and passengers, and has higher requirements on the production operation efficiency and the safety production management of the railway passenger station while more people use the railway passenger station as a first choice for traveling. The railway passenger station platform is a weak link, particularly, two ends of the platform belong to restricted areas in the station, and the passenger production management is influenced by the passenger mistakenly entering or intentionally entering. The 'high-speed railway safety protection and management method' passed in 2020 clearly indicates that warning signs and closed facilities should be installed and set at both ends of the platform to prevent irrelevant personnel from entering the high-speed railway line.
In order to ensure the normal boarding and alighting order of passengers and maintain the safety of restricted areas at two ends of a platform, most passenger stations adopt a person watching mode to solve the problem. But this entails an increase in personnel costs. With the development of the current advanced technology, a safety precaution mode combining technical defense and physical defense provides a new idea for the protection of two ends of the platform. There are many techniques for people detection, and the more common methods are: optical wave correlation, radar, image processing, etc. The light wave correlation technology adopts the alarm when the target blocks light, the alarm cannot be visualized, the interference of flying birds, fog, rain, snow and the like exists, and the detection angle is extremely narrow; the radar detection technology can realize a monitoring range of hundreds of meters, but the target resolving power is low, the visualization is poor, and whether workers can distinguish the targets or not can be further judged; and the implementation of the two technologies needs to install corresponding equipment, which brings higher construction cost and interferes the normal order of the passenger station in the construction process under the actual condition of the current passenger station. In the "high-speed railway safety protection management method", it is pointed out that a monitoring system is equipped and installed in important places such as a platform. The image detection technology can be selected on the basis of the original monitoring system, but the large-scale application of image processing is mainly the traditional method of artificial feature design, and although the traditional technology has a certain effect on pedestrian detection, the difficulty of distinguishing workers and passengers in pedestrians is still high. Although deep learning has the advantage of extracting and distinguishing object features, the complexity of operation of deep learning cannot meet the requirements of online application.
Disclosure of Invention
The invention aims to provide a non-worker intrusion detection and alarm method and device applied to a railway platform, so as to solve the problems in the prior art.
In a first aspect, the present invention provides a method for non-operator intrusion detection and alarm for a railway platform, comprising:
acquiring a real-time video of a station end area of a station platform;
detecting and tracking a target in the video by using an intrusion detection model, and judging whether to trigger an alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
Further, the intrusion detection model tracks the detected target by adopting a DCF algorithm, assigns a unique ID to the detected target, judges whether the targets in adjacent frames belong to the same target or not, and if the detected targets are the same, the ID assigned to the target is kept unchanged in the current frame.
Further, the determining whether to trigger an alarm based on the results of the detecting and tracking comprises:
defining a forbidden area for forbidding non-workers to enter from the real-time video;
determining to trigger an alarm when a target identified as a non-worker in the video is present in the forbidden zone and a target identified as a worker is absent in the effective range of the target identified as the non-worker; or
When the target identified as the non-staff member in the video appears in the forbidden zone and the target identified as the staff member exists in the effective range of the target identified as the non-staff member, an alarm is not triggered.
Further, the method further comprises: training a non-worker feature detection model based on YOLO v4, wherein training the non-worker feature detection model based on YOLO v4 comprises:
acquiring a video covering a station end forbidden area under a railway station scene, and extracting image frames in the video according to a preset rule;
constructing a data set comprising the image frames;
and carrying out transfer learning, model compression and reasoning acceleration on the pre-trained YOLO v4 network model by utilizing the data set.
Further, the performing migration learning on the pre-trained YOLO v4 network model by using the data set includes: and analyzing the data set, and supplementing or enhancing the unevenly distributed data.
Further, the model compression and reasoning acceleration are realized by using TensorRT and a network pruning method.
Further, the constructing a data set comprising the image frames comprises: and marking the extracted image frames based on the distinguishing features of the working personnel and the non-working personnel.
In a second aspect, the present invention provides a non-operator intrusion detection and alarm device for a railway platform, comprising:
the video acquisition module is used for acquiring a real-time video of a station end area of the platform;
the detection and alarm module is used for detecting and tracking the target in the video by using an intrusion detection model and judging whether to trigger alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff feature detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
In a third aspect, the present invention provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method for non-operator intrusion detection and alarm applied to a railway platform according to the first aspect when executing the program.
In a fourth aspect, the present invention provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, carries out the steps of the method for non-crew intrusion detection and alarm applied to a railway platform according to the first aspect.
According to the non-worker intrusion detection and alarm method and device applied to the railway station platform, provided by the invention, workers and non-workers can be distinguished by using an intrusion detection model, the real-time online analysis of dozens of paths of high-definition videos operated by one graphics processor can be realized, the false alarm caused by regular inspection of the workers is solved to a certain extent, the pressure of safety management of the passenger station when the passenger flow increases is effectively relieved, and the system is strong in reliability and low in cost.
Drawings
FIG. 1 is a flow chart of a non-operator intrusion detection and alarm method for a railway platform according to an embodiment of the present invention;
FIG. 2 is a flow chart of the training of a non-worker classification model based on YOLO v4 according to an embodiment of the present invention;
FIG. 3 is another flow chart of a non-operator intrusion detection and alarm method for a railway platform according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of a non-operator intrusion detection and alarm device for a railway platform according to an embodiment of the present invention; and
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The passenger flow of the high-speed railway passenger station is increased year by year, new requirements are provided for safety production management of the passenger station, particularly safety protection of two ends of a platform, the detection technology of radar and light wave correlation cannot solve the contradiction that workers routinely checking two ends of the platform (hereinafter, referred to as the station end) and mistakenly-entered passengers are effectively distinguished, and new equipment installed in the original environment is damaged. The traditional image processing technology has the problems that the extraction problem of non-worker recognition features cannot be solved, and the deep learning technology has complex operation and cannot be applied in a large scale. The invention aims to provide a station-end non-worker real-time detection method and an alarm system based on the combination of YOLO v4 and a Discriminant Correlation Filter (DCF) algorithm, aiming at the contradiction, so that the protection efficiency of a station-end area and the safety production management level of a passenger station can be greatly improved.
Specifically, the present invention provides a non-operator intrusion detection and alarm method applied to a railway platform, fig. 1 is a flowchart of a non-operator intrusion detection and alarm method applied to a railway platform according to an embodiment of the present invention, and referring to fig. 1, the non-operator intrusion detection and alarm method applied to a railway platform according to an embodiment of the present invention includes:
In the embodiment of the present invention, it should be noted that the intrusion detection model tracks a detected target by using a DCF algorithm, assigns a unique ID to the detected target, and determines whether targets in adjacent frames belong to the same target, and if the detected targets are the same, the ID assigned to the target remains unchanged in the current frame.
In particular, multi-target tracking using the DCF algorithm is further described below. On the premise of ensuring the detection speed of the algorithm, in order to further improve the robustness and feasibility of the algorithm, the intrusion detection model of the embodiment of the invention adopts a PtDCF algorithm, and relates to the tracking of multiple targets in the current frame. And judging whether each target detected in the adjacent frames belongs to the same object by using a DCF algorithm, wherein if the detected targets are the same, the ID allocated to the target is kept unchanged in the current frame. Compared with the traditional algorithms such as IOU, KLT and the like, the PtDCF algorithm adopted by the intrusion detection model provided by the embodiment of the invention has high precision, less resource consumption and high operation speed compared with deep learning associated algorithms such as deep learning.
In this embodiment of the present invention, it should be noted that the determining whether to trigger an alarm based on the detection and tracking result includes:
defining a forbidden area for forbidding non-workers to enter from the real-time video;
determining to trigger an alarm when a target identified as a non-worker in the video is present in the forbidden zone and a target identified as a worker is absent in the effective range of the target identified as the non-worker; or
When the target identified as the non-staff member in the video appears in the forbidden zone and the target identified as the staff member exists in the effective range of the target identified as the non-staff member, an alarm is not triggered.
In particular, the design of the alarm strategy is further elucidated below. The alarm strategy comprises the following steps:
and (1) defining a forbidden area. The station-end camera has a large field of view, and a forbidden region R needing to be monitored needs to be defined in a picture of the station-end camera;
and (2) carrying out synchronous real-time analysis on multiple cameras by using the YOLO network, and detecting the object in the current picture. Tracking the detected target by using a DCF algorithm, and allocating a unique ID for the target;
step (3), in the tracking process, if only a pedestrian P appears in a forbidden area, it indicates that a non-worker breaks into the area, immediately sending an alarm message (the format of the message is not fixed, but the content at least comprises the alarm type and the sending time) to the platform through socket communication (or other protocols and message middleware such as http, rabbitMQ and the like), and intercepting a video segment (the video duration is greater than or equal to 10s) to store and retain the evidence; if the pedestrian is detected to appear, and other labeled objects L exist in the adjacent area (the pixel point does not exceed 20) nearby, the situation that the object L is a worker or the worker exists in the effective range is indicated, and no alarm message is generated;
and (4) if the target ID is not changed all the time, no alarm message is generated under the current condition. If a new tracking ID is present, repeating steps (2) and (3).
In the embodiment of the present invention, it should be noted that the method further includes: training a non-worker feature detection model based on YOLO v4, wherein training the non-worker feature detection model based on YOLO v4 comprises:
acquiring a video covering a station end forbidden area, and extracting image frames in the video according to a preset rule;
constructing a data set comprising the image frames;
and carrying out transfer learning, model compression and reasoning acceleration on the pre-trained YOLO v4 network model by utilizing the data set.
Wherein the performing transfer learning on the pre-trained YOLO v4 network model by using the data set comprises: and analyzing the data set, and supplementing or enhancing the unevenly distributed data. The model compression and reasoning acceleration are realized by using TensorRT and a network pruning method. Said constructing a data set comprising said image frames comprises: and marking the extracted image frames based on the distinguishing features of the working personnel and the non-working personnel.
Specifically, fig. 2 is a flowchart of training a non-worker classification model based on YOLO v4 according to an embodiment of the present invention, and with reference to fig. 2, a method for training a non-worker classification model based on YOLO v4 according to an embodiment of the present invention is further described.
First a data set needs to be constructed. Collecting a camera video covering a station end forbidden area, wherein the content of the video comprises time spans of day and night in four seasons and different seasons, and extracting key frames or every n (n >5) frames of images in the video; the labeled categories in the dataset include categories for distinguishing workers from non-workers. And (3) marking the extracted image frames by using a marking tool (such as a LableiMG toolkit) to prepare a data set for training the YOLO, wherein the marked classification is not limited to railway spring and autumn clothing (including a hat), railway summer clothing (including a hat), railway winter clothing (including a hat), red and yellow reflective waistcoats, pedestrians, safety helmets, mops, cleaning trolleys and the like. A data set containing 3 ten thousand pictures is finally obtained.
Then, migration learning, model compression and reasoning acceleration are needed for the YOLO v4 network. Carrying out transfer learning on a pre-trained YOLO v4 network model by utilizing the constructed data set, wherein training parameters are not fixed, but in order to realize a better detection effect in the scene, the training parameters are proposed to be finely adjusted for many times, the constructed data set is analyzed, and unevenly distributed data are supplemented or data are enhanced; and (3) carrying out reasoning acceleration and model compression on the trained YOLO model by using a TensorRT and network pruning method to generate a new model YOLO _ Infer which can be deployed on GPU hardware. The model can reach at least 50FPS by using a T4 inference card.
Fig. 3 is another flowchart of a non-operator intrusion detection and alarm method applied to a railway platform according to an embodiment of the present invention, and the non-operator intrusion detection and alarm method applied to a railway platform according to an embodiment of the present invention will be further explained with reference to fig. 3.
Firstly, a monitoring system with a view field covering the areas at two ends of a railway platform acquires a real-time video. Wired or wireless surveillance cameras may be used, configured to comply with various safety regulations, and in particular, existing installed surveillance systems may be used.
And inputting the RTSP flow or the mixed flow obtained by the monitoring system into a YOLO _ Infer model. The YOLO _ inference model is obtained by a training method of a non-worker classification model based on YOLO v4 provided by the embodiment of the invention.
And then, tracking a target on the platform by adopting a DCF algorithm, and judging whether to trigger an alarm logic by using the alarm strategy provided by the embodiment of the invention. When the alarm logic is triggered, an alarm message is immediately sent to the platform via socket communications (or other protocol and message middleware such as http, rabbitMQ, etc.). When the alarm logic is not triggered, the detection and the tracking are continuously carried out through the intrusion detection model provided by the embodiment of the invention.
The invention belongs to the field of railway passenger station safety monitoring management under the application of a video intelligent analysis technology, overcomes the defects that the detection range of non-visible light technologies such as radar and light wave correlation is limited and workers and non-workers cannot be distinguished, and effectively overcomes the problems that the traditional video processing technology has poor identification performance and pain points caused by difficulty in artificial feature design and extraction and the deep learning model has high calculation force requirement and cannot be applied in a large scale in real time. The invention can realize the detection of the invasion of non-workers in the restricted areas at two ends of the railway station platform on the basis of the existing station monitoring system, can realize the real-time online analysis of dozens of paths of high-definition videos, can also solve the false alarm caused by the regular inspection of workers to a certain extent, and simultaneously effectively relieves the pressure of the safety management of the passenger station when the passenger flow increases, and has strong system reliability and low cost.
Referring to fig. 4, fig. 4 is a schematic structural diagram of a non-operator intrusion detection and alarm device applied to a railway platform according to an embodiment of the present invention, and the non-operator intrusion detection and alarm device applied to a railway platform provided in this embodiment includes: video acquisition module 410 and detection and alarm module 420:
a video obtaining module 410, configured to obtain a real-time video of a station end area of a station;
a detection and alarm module 420, configured to detect and track a target in the video using an intrusion detection model, and determine whether to trigger an alarm based on a result of the detection and tracking; the intrusion detection model detects a target in a video by using a non-staff feature detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
The non-operator intrusion detection and alarm device applied to the railway platform provided by the embodiment of the invention can be used for executing the non-operator intrusion detection and alarm method applied to the railway platform in the embodiment, and the working principle and the beneficial effect are similar, so detailed description is not provided here, and specific contents can be referred to the introduction of the embodiment.
In this embodiment, it should be noted that each module in the apparatus according to the embodiment of the present invention may be integrated into a whole or may be separately disposed. The modules can be combined into one module, and can also be further split into a plurality of sub-modules.
Fig. 5 illustrates a physical structure diagram of an electronic device, which may include, as shown in fig. 5: a processor (processor)510, a communication Interface (Communications Interface)520, a memory (memory)530 and a communication bus 540, wherein the processor 510, the communication Interface 520 and the memory 530 communicate with each other via the communication bus 540. Processor 510 may invoke logic instructions in memory 530 to perform a non-crew intrusion detection and alarm method for a railway platform, the method including obtaining real-time video of a station-side area of the platform; detecting and tracking a target in the video by using an intrusion detection model, and judging whether to trigger an alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff feature detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
Furthermore, the logic instructions in the memory 530 may be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the method for non-human intrusion detection and alarm applied to a railway platform, provided by the above methods, the method comprising: acquiring a real-time video of a station end area of a station platform; detecting and tracking a target in the video by using an intrusion detection model, and judging whether to trigger an alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff feature detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program, which when executed by a processor, implements the method for non-crew intrusion detection and alarm applied to a railway platform as provided above, the method comprising: acquiring a real-time video of a station end area of a station platform; detecting and tracking a target in the video by using an intrusion detection model, and judging whether to trigger an alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff feature detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A non-operator intrusion detection and alarm method applied to a railway platform is characterized by comprising the following steps:
acquiring a real-time video of a station end area of a station platform;
detecting and tracking a target in the video by using an intrusion detection model, and judging whether to trigger an alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff feature detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
2. The method of claim 1, wherein the intrusion detection model uses a DCF algorithm to track the detected objects and assign unique IDs to the detected objects, and determines whether the objects in adjacent frames belong to the same object, and if the detected objects are the same, the assigned IDs of the objects remain unchanged in the current frame.
3. The method of claim 1, wherein determining whether to trigger an alarm based on the results of the detecting and tracking comprises:
defining a forbidden area for forbidding non-workers to enter from the real-time video;
determining to trigger an alarm when a target identified as a non-worker in the video is present in the forbidden zone and a target identified as a worker is absent in the effective range of the target identified as the non-worker; or
When the target identified as the non-staff member in the video appears in the forbidden zone and the target identified as the staff member exists in the effective range of the target identified as the non-staff member, an alarm is not triggered.
4. The method of claim 1, further comprising: training a non-worker feature detection model based on YOLO v4, wherein training the non-worker feature detection model based on YOLO v4 comprises:
acquiring a video covering a station end forbidden area under a railway station scene, and extracting image frames in the video according to a preset rule;
constructing a data set comprising the image frames;
and carrying out transfer learning, model compression and reasoning acceleration on the pre-trained YOLO v4 network model by utilizing the data set.
5. The method of claim 4, wherein the using the data set to perform migration learning on the pre-trained YOLO v4 network model comprises: and analyzing the data set, and supplementing or enhancing the unevenly distributed data.
6. The non-human intrusion detection and alarm method for railway platforms according to claim 4 wherein the model compression and acceleration of inference is performed using TensorRT and web pruning methods.
7. The method of claim 4, wherein the constructing a data set including the image frames comprises: and marking the extracted image frames based on the distinguishing features of the working personnel and the non-working personnel.
8. A non-operator intrusion detection and alarm device for a railway platform, comprising:
the video acquisition module is used for acquiring a real-time video of a station end area of the platform;
the detection and alarm module is used for detecting and tracking the target in the video by using an intrusion detection model and judging whether to trigger alarm or not based on the detection and tracking result; the intrusion detection model detects a target in a video by using a non-staff detection model based on YOLO v4, and tracks the detected target by using a DCF algorithm.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the steps of the method of non-human intrusion detection and alarm for a railway platform according to any one of claims 1 to 7 when executing the program.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the method for non-human intrusion detection and alarm for a railway platform according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110751325.3A CN113537007A (en) | 2021-07-02 | 2021-07-02 | Non-worker intrusion detection and alarm method and device applied to railway platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110751325.3A CN113537007A (en) | 2021-07-02 | 2021-07-02 | Non-worker intrusion detection and alarm method and device applied to railway platform |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113537007A true CN113537007A (en) | 2021-10-22 |
Family
ID=78126584
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110751325.3A Pending CN113537007A (en) | 2021-07-02 | 2021-07-02 | Non-worker intrusion detection and alarm method and device applied to railway platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113537007A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114283544A (en) * | 2021-12-10 | 2022-04-05 | 中国电子科技集团公司第三十八研究所 | Railway platform intrusion monitoring system and method based on artificial intelligence |
| CN116597587A (en) * | 2023-05-31 | 2023-08-15 | 河南龙宇能源股份有限公司 | Underground operation equipment high-risk area invasion early warning method based on audio-visual cooperative recognition |
Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101635835A (en) * | 2008-07-25 | 2010-01-27 | 深圳市信义科技有限公司 | Intelligent video monitoring method and system thereof |
| CN102013147A (en) * | 2010-09-29 | 2011-04-13 | 北京航空航天大学 | Intelligent burglary prevention monitoring method and device for high-voltage power transmission tower |
| CN106507037A (en) * | 2016-09-30 | 2017-03-15 | 北京中星微电子有限公司 | Intelligent control method and system that personnel invade/cross the border |
| CN108765817A (en) * | 2018-08-17 | 2018-11-06 | 西南交大(上海)智能***有限公司 | A kind of railroad platform intrusion alarm system |
| CN109040669A (en) * | 2018-06-28 | 2018-12-18 | 国网山东省电力公司菏泽供电公司 | Intelligent substation video fence method and system |
| CN109448116A (en) * | 2018-10-31 | 2019-03-08 | 广西路桥工程集团有限公司 | A kind of wisdom fielded system based on BIM model |
| CN109920186A (en) * | 2019-04-19 | 2019-06-21 | 沈阳风驰软件股份有限公司 | A kind of detection of platform edge and geofence control system and method |
| US20190325584A1 (en) * | 2018-04-18 | 2019-10-24 | Tg-17, Llc | Systems and Methods for Real-Time Adjustment of Neural Networks for Autonomous Tracking and Localization of Moving Subject |
| CN110532852A (en) * | 2019-07-09 | 2019-12-03 | 长沙理工大学 | Subway station pedestrian's accident detection method based on deep learning |
| CN110675586A (en) * | 2019-09-25 | 2020-01-10 | 捻果科技(深圳)有限公司 | Airport enclosure intrusion monitoring method based on video analysis and deep learning |
| CN110781964A (en) * | 2019-10-28 | 2020-02-11 | 兰州交通大学 | Human body target detection method and system based on video image |
| CN110796819A (en) * | 2019-10-18 | 2020-02-14 | 中国铁道科学研究院集团有限公司电子计算技术研究所 | Detection method and system for platform yellow line invasion border crossing personnel |
| CN111192426A (en) * | 2020-01-14 | 2020-05-22 | 中兴飞流信息科技有限公司 | Railway perimeter intrusion detection method based on anthropomorphic visual image analysis video cruising |
| CN111310934A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | Model generation method and device, electronic equipment and storage medium |
| CN111832457A (en) * | 2020-07-01 | 2020-10-27 | 济南浪潮高新科技投资发展有限公司 | Stranger intrusion detection method based on cloud edge cooperation |
| CN112052824A (en) * | 2020-09-18 | 2020-12-08 | 广州瀚信通信科技股份有限公司 | Gas pipeline specific object target detection alarm method, device and system based on YOLOv3 algorithm and storage medium |
| CN112257492A (en) * | 2020-08-27 | 2021-01-22 | 重庆科技学院 | Real-time intrusion detection and tracking method for multiple cameras |
| CN112381778A (en) * | 2020-11-10 | 2021-02-19 | 国网浙江嵊州市供电有限公司 | Transformer substation safety control platform based on deep learning |
| CN112434828A (en) * | 2020-11-23 | 2021-03-02 | 南京富岛软件有限公司 | Intelligent identification method for safety protection in 5T operation and maintenance |
| CN113034541A (en) * | 2021-02-26 | 2021-06-25 | 北京国双科技有限公司 | Target tracking method and device, computer equipment and storage medium |
-
2021
- 2021-07-02 CN CN202110751325.3A patent/CN113537007A/en active Pending
Patent Citations (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101635835A (en) * | 2008-07-25 | 2010-01-27 | 深圳市信义科技有限公司 | Intelligent video monitoring method and system thereof |
| CN102013147A (en) * | 2010-09-29 | 2011-04-13 | 北京航空航天大学 | Intelligent burglary prevention monitoring method and device for high-voltage power transmission tower |
| CN106507037A (en) * | 2016-09-30 | 2017-03-15 | 北京中星微电子有限公司 | Intelligent control method and system that personnel invade/cross the border |
| US20190325584A1 (en) * | 2018-04-18 | 2019-10-24 | Tg-17, Llc | Systems and Methods for Real-Time Adjustment of Neural Networks for Autonomous Tracking and Localization of Moving Subject |
| CN109040669A (en) * | 2018-06-28 | 2018-12-18 | 国网山东省电力公司菏泽供电公司 | Intelligent substation video fence method and system |
| CN108765817A (en) * | 2018-08-17 | 2018-11-06 | 西南交大(上海)智能***有限公司 | A kind of railroad platform intrusion alarm system |
| CN109448116A (en) * | 2018-10-31 | 2019-03-08 | 广西路桥工程集团有限公司 | A kind of wisdom fielded system based on BIM model |
| CN109920186A (en) * | 2019-04-19 | 2019-06-21 | 沈阳风驰软件股份有限公司 | A kind of detection of platform edge and geofence control system and method |
| CN110532852A (en) * | 2019-07-09 | 2019-12-03 | 长沙理工大学 | Subway station pedestrian's accident detection method based on deep learning |
| CN110675586A (en) * | 2019-09-25 | 2020-01-10 | 捻果科技(深圳)有限公司 | Airport enclosure intrusion monitoring method based on video analysis and deep learning |
| CN110796819A (en) * | 2019-10-18 | 2020-02-14 | 中国铁道科学研究院集团有限公司电子计算技术研究所 | Detection method and system for platform yellow line invasion border crossing personnel |
| CN110781964A (en) * | 2019-10-28 | 2020-02-11 | 兰州交通大学 | Human body target detection method and system based on video image |
| CN111192426A (en) * | 2020-01-14 | 2020-05-22 | 中兴飞流信息科技有限公司 | Railway perimeter intrusion detection method based on anthropomorphic visual image analysis video cruising |
| CN111310934A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | Model generation method and device, electronic equipment and storage medium |
| CN111832457A (en) * | 2020-07-01 | 2020-10-27 | 济南浪潮高新科技投资发展有限公司 | Stranger intrusion detection method based on cloud edge cooperation |
| CN112257492A (en) * | 2020-08-27 | 2021-01-22 | 重庆科技学院 | Real-time intrusion detection and tracking method for multiple cameras |
| CN112052824A (en) * | 2020-09-18 | 2020-12-08 | 广州瀚信通信科技股份有限公司 | Gas pipeline specific object target detection alarm method, device and system based on YOLOv3 algorithm and storage medium |
| CN112381778A (en) * | 2020-11-10 | 2021-02-19 | 国网浙江嵊州市供电有限公司 | Transformer substation safety control platform based on deep learning |
| CN112434828A (en) * | 2020-11-23 | 2021-03-02 | 南京富岛软件有限公司 | Intelligent identification method for safety protection in 5T operation and maintenance |
| CN113034541A (en) * | 2021-02-26 | 2021-06-25 | 北京国双科技有限公司 | Target tracking method and device, computer equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| Y LIU等: ""Non-staff Detection for End-Intrusion in Railway Station Platform Based on YOLO and DCF Techniques"", 《HTTPS://LINK.SPRINGER.COM/CHAPTER/10.1007/978-981-16-9909-2_20》, 19 February 2022 (2022-02-19) * |
| ZHIJIAN QU等: "全文", 《IEEE ACCESS 》, 4 September 2019 (2019-09-04) * |
| 王瑞: ""一种基于视频的铁路周界入侵检测智能综合识别技术研究"", 《仪器仪表学报》, 15 September 2020 (2020-09-15), pages 1 - 4 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114283544A (en) * | 2021-12-10 | 2022-04-05 | 中国电子科技集团公司第三十八研究所 | Railway platform intrusion monitoring system and method based on artificial intelligence |
| CN116597587A (en) * | 2023-05-31 | 2023-08-15 | 河南龙宇能源股份有限公司 | Underground operation equipment high-risk area invasion early warning method based on audio-visual cooperative recognition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103106766B (en) | Forest fire identification method and forest fire identification system | |
| CN108062349B (en) | Video monitoring method and system based on video structured data and deep learning | |
| CN105744232B (en) | A kind of method of the transmission line of electricity video external force damage prevention of Behavior-based control analytical technology | |
| KR102122859B1 (en) | Method for tracking multi target in traffic image-monitoring-system | |
| CN103069434B (en) | For the method and system of multi-mode video case index | |
| CN103108159B (en) | Electric power intelligent video analyzing and monitoring system and method | |
| CN110032977A (en) | A kind of safety warning management system based on deep learning image fire identification | |
| CN113537007A (en) | Non-worker intrusion detection and alarm method and device applied to railway platform | |
| KR20200058260A (en) | Apparatus for CCTV Video Analytics Based on Object-Image Recognition DCNN and Driving Method Thereof | |
| CN106741008B (en) | Railway line foreign matter identification method and system | |
| CN113469654B (en) | Multi-level safety control system of transformer substation based on intelligent algorithm fuses | |
| KR102122850B1 (en) | Solution for analysis road and recognition vehicle license plate employing deep-learning | |
| CN104981818A (en) | Systems and methods to classify moving airplanes in airports | |
| Ozcelik et al. | A vision based traffic light detection and recognition approach for intelligent vehicles | |
| CN111753651A (en) | Subway group abnormal behavior detection method based on station two-dimensional crowd density analysis | |
| CN111931726B (en) | Traffic light detection method, device, computer storage medium and road side equipment | |
| CN110263623B (en) | Train climbing monitoring method, device, terminal and storage medium | |
| CN111918039A (en) | Artificial intelligence high risk operation management and control system based on 5G network | |
| CN106412508A (en) | Intelligent monitoring method and system of illegal line press of vehicles | |
| CN112287823A (en) | Facial mask identification method based on video monitoring | |
| CN110781844A (en) | Security patrol monitoring method and device | |
| CN102324018A (en) | Pedestrian safety state recognition method and system of comprehensive transportation interchange service network | |
| CN111985295A (en) | Electric bicycle behavior recognition method and system, industrial personal computer and camera | |
| CN113283273B (en) | Method and system for detecting front obstacle in real time based on vision technology | |
| CN112686130A (en) | Wisdom fishing boat supervision decision-making system |
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 |