CN115376037A - Station key area safety state monitoring method based on video - Google Patents
Station key area safety state monitoring method based on video Download PDFInfo
- Publication number
- CN115376037A CN115376037A CN202210815091.9A CN202210815091A CN115376037A CN 115376037 A CN115376037 A CN 115376037A CN 202210815091 A CN202210815091 A CN 202210815091A CN 115376037 A CN115376037 A CN 115376037A
- Authority
- CN
- China
- Prior art keywords
- frame
- passenger flow
- target
- tracking
- target detection
- 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
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- 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
-
- 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]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The invention provides a station key area safety state monitoring method based on videos. The method comprises the following steps: acquiring a video image in a monitoring area under a section of track traffic station scene, and performing passenger flow target detection on the video image by using a target detection algorithm to acquire a passenger flow target detection frame; carrying out target tracking on the passenger flow target detection frame by using an improved DaSim-RPN algorithm to obtain a passenger flow target tracking frame of the current frame; judging whether the track of the passenger flow target tracking frame of the current frame exceeds the monitoring area, and if so, stopping the target tracking process; otherwise, continuing the target tracking process of the video image. According to the invention, a deep learning method is applied to a passenger flow information detection scene of a rail transit station, and the YOLOv3 and an improved DaSim-RPN algorithm are utilized to improve the passenger flow information monitoring precision, so that the target monitoring under the shielding condition can be effectively improved, the running speed is improved, and the real-time performance of the passenger flow information detection is ensured.
Description
Technical Field
The invention relates to the technical field of rail transit video monitoring, in particular to a method for monitoring the safety state of a station key area based on videos.
Background
The normal operation of the rail transit station has great significance to the operation safety of the whole urban rail transit network, and the accurate grasp of the passenger flow operation state of the station is the fundamental basis for realizing the scientific management and control of the passenger flow.
At present, a neural network method is mostly adopted in a video monitoring method in the prior art, the method is widely applied to a rail transit station scene, but is still limited by various factors of field engineering, the passenger flow state information monitoring in a specific area in a station cannot achieve a good effect, and the method cannot be known at first time when dangerous states such as large passenger flow congestion, channel retrograde motion and non-working personnel entering an operation area occur.
Therefore, how to develop an effective video-based station key area safety state monitoring method is an urgent problem to be solved.
Disclosure of Invention
The embodiment of the invention provides a method for monitoring the safety state of a station key area based on videos, so as to effectively monitor the safety of the station key area.
In order to achieve the purpose, the invention adopts the following technical scheme.
A safety state monitoring method for a station key area based on videos comprises the following steps:
acquiring a video image in a monitoring area under a section of track traffic station scene, and performing passenger flow target detection on the video image by using a target detection algorithm to acquire a passenger flow target detection frame;
carrying out target tracking on the passenger flow target detection frame by using an improved DaSim-RPN algorithm to obtain a passenger flow target tracking frame of the current frame;
judging whether the track of the passenger flow target tracking frame of the current frame exceeds the monitoring area, and if so, stopping the target tracking process; and if not, continuing the target tracking process on the video image.
Preferably, the method comprises the steps of obtaining a video image in a monitoring area under a section of track traffic station scene, carrying out passenger flow target detection on the video image by using a target detection algorithm, and obtaining a passenger flow target detection frame;
acquiring video images of passage passenger flows in a monitoring area under a section of rail transit station scene, and extracting and storing continuous frames of the video images;
and carrying out passenger flow target detection on the continuous frames of the video image by using a target detection algorithm to obtain a passenger flow target detection frame and coordinates and confidence coefficients of four vertexes of the detection frame.
Preferably, the target detection algorithm comprises a deep learning-based YOLOv3 algorithm.
Preferably, the improved DaSim-RPN algorithm is used for carrying out target tracking on the passenger flow target detection frame to obtain a passenger flow target tracking frame of the current frame, and the method comprises the following steps of;
screening the detected passenger flow target detection frames, and establishing initial tracking positions corresponding to the targets by using the screened detection frames;
in the first frame picture, a region of 127 × 127 is intercepted from the central point of the initial tracking target and is used as a template frame region; in the subsequent pictures, a 255-255 region is intercepted from the central point of the tracking target in the previous frame and is used as a search frame region;
finding k candidate region frames closest to the tracking target central point O, calculating the cross-over ratio IoU of each candidate region frame, calculating the mean value and the variance of the k IoUs, setting the mean value plus the variance as an adaptive threshold, defining the candidate region frames larger than the adaptive threshold as interferents, removing the interferents, and forming all the candidate region frames with the interferents removed into a search frame region;
template frame area andthe search frame area is sent into an improved DaSim-RPN network, the improved DaSim-RPN network outputs a tracking frame and a score of a target, the score is subjected to cosine window suppression, the score of an edge area far away from a central point is punished, the central point of the tracking frame with the highest score is taken as a new central point (A'), the width W height H of the target of the previous frame and the width W of the tracking frame are taken as g High H g And performing smooth weighting to obtain a new width W ' height H ', adopting a new central point A ' and the new width W ' height H ' as passenger flow target tracking frames of the current frame, and performing subsequent tracking and loop iteration in sequence.
Preferably, the method comprises the steps of judging whether the track of the passenger flow target tracking frame of the current frame exceeds the monitoring area, and if so, stopping the target tracking process; if not, continuing to perform a target tracking process on the video image, including;
judging whether the track of the passenger flow target tracking frame of the previous frame exceeds a defined monitoring area or not, and stopping the target tracking process if the track of the passenger flow target tracking frame of the previous frame exceeds the defined monitoring area; and if not, continuing the target tracking process to monitor the passenger flow safety.
According to the technical scheme provided by the embodiment of the invention, the deep learning method is applied to the passenger flow information detection scene of the rail transit station, the YOLOv3 and the improved DaSim-RPN algorithm are utilized to improve the passenger flow information monitoring precision, the target monitoring under the shielding condition can be effectively improved, the running speed is increased, and the real-time performance of the passenger flow information detection is ensured. The invention can well monitor the designated area in the station and provides effective guarantee for the passenger flow and the environmental safety of the target area.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the description below are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a schematic diagram illustrating an implementation of a method for monitoring a safety state of a station key area based on video according to an embodiment of the present invention;
fig. 2 is a flowchart of a method for monitoring safety status of a station key area based on video according to an embodiment of the present invention;
fig. 3 is a schematic view of an aisle experiment scene of a rail transit station according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or coupled. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
For the convenience of understanding the embodiments of the present invention, the following description will be further explained by taking several specific embodiments as examples in conjunction with the drawings, and the embodiments are not to be construed as limiting the embodiments of the present invention.
The embodiment of the invention provides a method for monitoring the safety state of a station key area based on a video, which solves the problem of monitoring a user-defined area in a rail transit station scene by combining YOLOv3 and an improved DaSim-RPN algorithm.
An implementation schematic diagram of the station key area safety state monitoring method based on the video is shown in fig. 1, and a specific processing flow is shown in fig. 2, and the method includes the following processing steps:
and S10, acquiring a video image in a monitoring area in a section of track traffic station scene, and performing passenger flow target detection on the video image by using a target detection algorithm to acquire a passenger flow target detection frame.
The method comprises the steps of obtaining video images of passage passenger flow in a monitoring area under a section of rail transit station scene, extracting and storing continuous frames of the video images, and extracting 24 frames every 1 s.
And carrying out passenger flow target detection on the continuous frames of the video image by using a target detection algorithm to obtain a passenger flow target detection frame and coordinates and confidence coefficients of four vertexes of the detection frame. The target detection algorithm may be a deep learning-based YOLOv3 algorithm.
And step S20, carrying out target tracking on the passenger flow target detection frame by using an improved DaSim-RPN algorithm, and acquiring a passenger flow target tracking frame of the current frame.
And screening the detected passenger flow target detection frames, and establishing an initial tracking position corresponding to the target by using the screened detection frames, namely an initial tracking target.
In the first frame picture, a region of 127 × 127 is cut out from a central point (A) of an initial tracking target and is used as a template frame region; in the subsequent pictures, a 255 × 255 region is cut from the central point (B) of the tracking target of the previous frame, and the region is used as a search frame region.
K candidate region frames closest to the tracking target central point O, and calculating IoU of each candidate region frame
(Intersection overlapping unit), then, calculating the mean and variance of k IoUs, setting the mean plus variance as an Adaptive threshold by an ATSS (Adaptive Training Sample Selection) method, finally defining some candidate region frames larger than the Adaptive threshold as interferents, and removing the interferents. And forming a search frame region by all the candidate region frames from which the interferents are removed.
And sending a template frame region (namely a real image of the first frame) and the search frame region into a modified DaSiam-RPN network, wherein the modified DaSiam-RPN network predicts a tracking frame and a score of a target, and the template frame region is the real image of the first frame.
The improved DaSim-RPN network carries out similarity calculation on the template frame region target and the search frame region target, so that a tracking target which is most similar to a tracking target in a template frame is searched in the search frame region, and a new incremental learning method is introduced, so that the tracking frame prediction is more stable, the tracking ID jump is reduced, and a corresponding tracking frame and a corresponding score are predicted.
Performing cosine window suppression on the score, namely punishing the score of an edge area far away from a central point (A), taking the central point of a tracking frame with the highest score as a new central point (A'), and taking the width (W) and the height (H) of a target in the previous frame and the width (W) of the tracking frame g ) High (H) g ) The smooth weighting is performed as the new width (W ') height (H').
Adopting the new central point (A ') and the width (W ') and the height (H ') as passenger flow target tracking frames of the current frame, and sequentially and circularly iterating the subsequent tracking; and setting the weight factor of the passenger flow target tracking frame as 1/2 of the weight of the interference object. And in the proportion setting in the new increment learning, the weight factor of the tracking frame is set to be 1/2 of the weight of the interference object, so that the tracking target is more stable and the ID jump does not occur.
Step S30, judging whether the track of the latest passenger flow target tracking frame exceeds a defined monitoring area, and if so, stopping the target tracking process; and if not, continuing the target tracking process to monitor the passenger flow safety.
In order to verify the effectiveness and the accuracy of the invention, a certain passage scene of the rail transit station is selected for testing, as shown in fig. 3, a solid line in fig. 3 is a limit solid line which is manually calibrated, when a tracked target crosses the limit solid line, the tracking ID of the tracked target is recorded, and therefore, the cross-border people counting and the like can be carried out.
In summary, the embodiment of the invention provides a detection model based on deep learning, aiming at the problem of monitoring the safety state of the rail transit station in the key area. The model detects the passenger flow target through a YOLOv3 algorithm, and transmits the detection result to an improved DaSim-RPN algorithm in real time for target tracking. The target tracking missing report rate can be reduced, the tracking accuracy is improved, and the aim of monitoring the passenger flow in the designated area monitored by the video is well fulfilled.
The improved DaSim-RPN algorithm of the embodiment of the invention leads the threshold value to be more reasonably set by introducing a self-adaptive threshold value method, can further screen out interferents and inhibit the influence of the interferents, thereby generating a stable tracking frame; meanwhile, the corresponding relation between the weight factor and the weight of the interferent is changed, so that the weight factor is changed along with the change of the weight of the interferent, and the tracking precision of the target is improved. Compared with other deep learning methods, the method has the advantages of lower missing report rate and higher accuracy rate of monitoring the key area of the rail transit station, simultaneously improves the tracking efficiency, consumes less computer resources and is suitable for the field environment.
Those of ordinary skill in the art will understand that: the figures are merely schematic representations of one embodiment, and the blocks or flow diagrams in the figures are not necessarily required to practice the present invention.
From the above description of the embodiments, it is clear to those skilled in the art that the present invention can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored in a 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 method according to the embodiments or some parts of the embodiments.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for apparatus or system embodiments, since they are substantially similar to method embodiments, they are described in relative terms, as long as they are described in partial descriptions of method embodiments. The above-described embodiments of the apparatus and system are merely illustrative, and the units described as separate parts may or may not be physically separate, and the 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 this embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (5)
1. A safety state monitoring method for a station key area based on videos is characterized by comprising the following steps:
acquiring a video image in a monitoring area under a section of track traffic station scene, and performing passenger flow target detection on the video image by using a target detection algorithm to acquire a passenger flow target detection frame;
carrying out target tracking on the passenger flow target detection frame by using an improved DaSim-RPN algorithm to obtain a passenger flow target tracking frame of the current frame;
judging whether the track of the passenger flow target tracking frame of the current frame exceeds the monitoring area, and if so, stopping the target tracking process; and if not, continuing the target tracking process on the video image.
2. The method according to claim 1, wherein the step of obtaining the video image in the monitoring area under the scene of a section of track traffic station, using a target detection algorithm to perform passenger flow target detection on the video image, and obtaining a passenger flow target detection frame comprises;
acquiring video images of passage passenger flows in a monitoring area under a section of rail transit station scene, and extracting and storing continuous frames of the video images;
and carrying out passenger flow target detection on the continuous frames of the video image by using a target detection algorithm, and acquiring a passenger flow target detection frame and coordinates and confidence coefficients of four vertexes of the detection frame.
3. The method of claim 2, wherein the target detection algorithm comprises a deep learning based YOLOv3 algorithm.
4. The method according to claim 1, wherein the improved DaSiam-RPN algorithm is used for carrying out target tracking on the passenger flow target detection frame to obtain a passenger flow target tracking frame of the current frame, including;
screening the detected passenger flow target detection frames, and establishing initial tracking positions corresponding to the targets by using the screened detection frames;
in the first frame picture, a region of 127 × 127 is intercepted from the central point of the initial tracking target and is used as a template frame region; in the subsequent pictures, a 255 x 255 region is intercepted from the central point of the tracking target of the previous frame and is used as a search frame region;
finding k candidate region frames closest to the tracking target central point O, calculating the cross-over ratio IoU of each candidate region frame, calculating the mean value and the variance of the k IoUs, setting the mean value plus the variance as an adaptive threshold, defining the candidate region frames larger than the adaptive threshold as interferents, removing the interferents, and forming all the candidate region frames with the interferents removed into a search frame region;
sending the template frame region and the search frame region into an improved DaSim-RPN network, outputting a tracking frame and a score of a target by the improved DaSim-RPN network, performing cosine window suppression on the score, punishing the score of an edge region far away from a central point, and taking the central point of the tracking frame with the highest score as a new central point (A'), wherein the width W and the height H of the target of the previous frame and the width W and the height H of the tracking frame g High H g And performing smooth weighting to obtain a new width W ' height H ', adopting a new central point A ' and the new width W ' height H ' as passenger flow target tracking frames of the current frame, and performing subsequent tracking and loop iteration in sequence.
5. The method according to claim 1, wherein the method comprises the steps of judging whether the track of the passenger flow target tracking frame of the current frame exceeds the monitoring area, and if so, stopping the target tracking process; if not, continuing to perform a target tracking process on the video image, including;
judging whether the track of the passenger flow target tracking frame of the previous frame exceeds a defined monitoring area or not, and if so, stopping the target tracking process; and if not, continuing the target tracking process to monitor the passenger flow safety.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210815091.9A CN115376037A (en) | 2022-07-12 | 2022-07-12 | Station key area safety state monitoring method based on video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210815091.9A CN115376037A (en) | 2022-07-12 | 2022-07-12 | Station key area safety state monitoring method based on video |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115376037A true CN115376037A (en) | 2022-11-22 |
Family
ID=84062196
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210815091.9A Pending CN115376037A (en) | 2022-07-12 | 2022-07-12 | Station key area safety state monitoring method based on video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115376037A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117423067A (en) * | 2023-12-18 | 2024-01-19 | 成都华芯智云科技有限公司 | Passenger flow statistics terminal based on TOF technology |
-
2022
- 2022-07-12 CN CN202210815091.9A patent/CN115376037A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117423067A (en) * | 2023-12-18 | 2024-01-19 | 成都华芯智云科技有限公司 | Passenger flow statistics terminal based on TOF technology |
| CN117423067B (en) * | 2023-12-18 | 2024-03-12 | 成都华芯智云科技有限公司 | Passenger flow statistics terminal based on TOF technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Aboah | A vision-based system for traffic anomaly detection using deep learning and decision trees | |
| US20210248378A1 (en) | Spatiotemporal action detection method | |
| CN102378992B (en) | Articulated region detection device and method for same | |
| CN111160125A (en) | Railway foreign matter intrusion detection method based on railway monitoring | |
| CN104966304A (en) | Kalman filtering and nonparametric background model-based multi-target detection tracking method | |
| Lee et al. | Real-time illegal parking detection in outdoor environments using 1-D transformation | |
| CN102073851A (en) | Method and system for automatically identifying urban traffic accident | |
| CN112560641A (en) | Video-based one-way passenger flow information detection method in two-way passenger flow channel | |
| Gong et al. | Local distinguishability aggrandizing network for human anomaly detection | |
| Chen et al. | Dual-modality vehicle anomaly detection via bilateral trajectory tracing | |
| CN114925747A (en) | Vehicle abnormal running detection method, electronic device, and storage medium | |
| CN108830204B (en) | Method for detecting abnormality in target-oriented surveillance video | |
| CN114898326A (en) | Method, system and equipment for detecting reverse running of one-way vehicle based on deep learning | |
| CN113673311A (en) | Traffic abnormal event detection method, equipment and computer storage medium | |
| CN115376037A (en) | Station key area safety state monitoring method based on video | |
| Saypadith et al. | An approach to detect anomaly in video using deep generative network | |
| CN112967320A (en) | Ship target detection tracking method based on bridge collision avoidance | |
| CN102789645A (en) | Multi-objective fast tracking method for perimeter precaution | |
| CN101877135A (en) | Moving target detecting method based on background reconstruction | |
| CN116311166A (en) | Traffic obstacle recognition method and device and electronic equipment | |
| CN115512263A (en) | Dynamic visual monitoring method and device for falling object | |
| Batapati et al. | Video analysis for traffic anomaly detection using support vector machines | |
| Ikoma et al. | Multi-target tracking in video by SMC-PHD filter with elimination of other targets and state dependent multi-modal likelihoods | |
| CN114640807A (en) | Video-based object counting method and device, electronic equipment and storage medium | |
| KR20220067271A (en) | Image acquisition apparatus and image acquisition method |
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 |