Surveillance video sharing method, surveillance video monitoring system and electronic equipment

Abstract

The invention discloses a monitoring video sharing method, a monitoring system thereof and electronic equipment, which are applied to the monitoring system consisting of a plurality of cameras and a cloud platform, the plurality of cameras are connected through a local area network, and the method comprises the following steps: detecting other cameras in the local area network by each camera in the plurality of cameras, the detected other camera information is reported to the cloud platform; the cloud platform allocates a corresponding algorithm model to each camera based on the scene selected by the user, wherein each algorithm model is configured to detect a specific target object or scene; each camera runs the distributed algorithm model; each camera sends video data collected by the camera to other cameras through a local area network and receives video data sent by other cameras at the same time; the camera processes the video data collected and received by the camera by using the running algorithm model to generate a detection result; and each camera sends a detection result to the cloud platform.

Description

Technical Field

[0001] The present invention relates to the field of smart security technology, and in particular to a monitoring video sharing method and a monitoring system and electronic equipment thereof. Background Art

[0002] In existing surveillance systems, multiple cameras are typically used to monitor different areas. The video data collected by these cameras often needs to be shared and collaboratively processed to achieve comprehensive security monitoring. However, traditional surveillance video sharing methods have some limitations.

[0003] On the one hand, some methods use a single-camera, single-algorithm model, where the camera runs a simple algorithm model locally, such as human detection. Due to the limited computing power on the camera side, it is impossible to run multiple complex algorithm models simultaneously to meet diverse monitoring needs in different scenarios, such as simultaneous human, vehicle, and flame detection. This results in a relatively limited monitoring system functionality, making it impossible to achieve multi-dimensional security monitoring.

[0004] On the other hand, while some approaches utilize a multi-camera and cloud-based multimodal computing architecture, this approach requires high cloud computing power, consumes significant cloud resources, and poses data leakage and compliance risks. Furthermore, approaches that utilize multiple cameras plus on-device edge computing gateways require the purchase of additional on-device edge computing gateways, increasing system hardware costs and also exposing the risk of data leakage from the edge computing gateways.

[0005] In summary, existing technologies have shortcomings in meeting multi-scenario monitoring needs, ensuring data security, and controlling hardware costs. A more efficient, secure, and economical surveillance video sharing method and system is needed to solve these problems. Summary of the Invention

[0006] In view of the problems existing in the prior art, the present invention provides a method for sharing surveillance videos.

[0007] To achieve the above object, the present invention provides a method for sharing surveillance videos, which is applied to a surveillance system consisting of multiple cameras and a cloud platform, wherein the multiple cameras are connected via a local area network (LAN). The method comprises the following steps: Step S1: Each of the multiple cameras detects other cameras in the local area network and reports the detected information of the other cameras to the cloud platform; Step S2: The cloud platform assigns a corresponding algorithm model to each camera based on the scene selected by the user, where each algorithm model is configured to detect a specific target object or scene; Step S3: Each camera runs its assigned algorithm model; Step S4: Each camera sends the video data collected by itself to other cameras through the local area network, and receives the video data sent by other cameras at the same time; Step S5: The camera uses the running algorithm model to process the video data collected and received by itself to generate a detection result; Step S6: Each camera sends the detection results to the cloud platform.

[0008] Preferably, the scenario selected by the user in step S2 includes at least one of human detection, vehicle detection or flame detection, which is used to determine the type of algorithm model assigned to the camera.

[0009] Preferably, the video data is sub-stream data with a lower resolution than the main stream, wherein each camera is divided into two streams, wherein the main stream resolution is 1920*1080 and the sub-stream resolution is 640*360.

[0010] Preferably, in step S4, the resolution video data of each sub-stream is encrypted and transmitted to other cameras through the RTSP protocol at the same time. After each camera receives the data through RTSP, it decrypts it and stores it in a ring buffer. After extracting frames to detect the video data, the output detection result is encapsulated in JSON format and sent to the cloud platform. The cloud platform then distributes the detection result to the corresponding user according to the camera ID of the detection result.

[0011] Preferably, the detection result includes: camera identification, timestamp, target coordinates and confidence level.

[0012] The present invention also provides a monitoring system, comprising: a plurality of cameras and a cloud platform; The multiple cameras are interconnected via a local area network, and each camera includes: an algorithm running module for executing the algorithm model assigned by the cloud platform; a video transmission module for sending and receiving sub-stream video data; a data encryption module for encrypting and decrypting transmitted data; and a result encapsulation module for encapsulating the detection results into JSON format data; The cloud platform is used to receive information about other cameras in the local area network reported by the camera, assign an algorithm model to the camera according to the scene selected by the user, receive detection results and distribute them to the corresponding users.

[0013] Preferably, the camera is configured with an NPU, and each NPU only runs a single algorithm model.

[0014] Preferably, the video transmission module supports TCP / IP or UDP protocol.

[0015] Preferably, the cloud platform further includes a decision-making module for performing fusion analysis on the detection results and distributing the results to different service terminals (such as APP or client) after making decisions.

[0016] The present invention also provides an electronic device, comprising a processor and a memory, wherein the memory stores a computer program, and the processor implements the steps of the sharing method when executing the computer program.

[0017] The technical solution of the present invention has the following beneficial effects: This invention reduces hardware costs: there is no need to deploy edge computing gateways, and equipment investment can be saved through distributed collaboration of cameras; This invention improves computing power utilization: each camera runs a single algorithm, and through data sharing, an "N×N" detection matrix is ​​formed (N is the number of cameras), avoiding repeated calculations; This invention reduces cloud load: by only uploading detection results (instead of original videos), the cloud computing power required for 100 cameras is reduced to 10% of the traditional solution; This invention optimizes privacy and real-time performance: raw video data is transmitted within the local area network, and sensitive information (such as faces) does not leave the local area network, complying with regulations such as GDPR; detection delay is <100ms, meeting real-time warning needs.

[0018] This invention enables efficient sharing and collaborative processing: it enables efficient video data sharing between multiple cameras, transmitting encrypted substream video data via a local area network, ensuring efficient data transmission while reducing network bandwidth requirements. Video information can be shared between cameras in real time, avoiding the problem of information silos. The cameras collaborate to run different algorithm models, forming an integrated monitoring network. For example, one camera can run a human detection algorithm while another runs a vehicle detection algorithm. These algorithms can work together to achieve comprehensive monitoring of complex scenes.

[0019] This invention provides intelligent monitoring: the cloud platform assigns corresponding algorithm models to cameras based on user-selected scenarios, enabling the monitoring system to flexibly adapt to diverse monitoring needs. Users can select algorithm models for human detection, vehicle detection, or flame detection based on the specific scenario, enhancing the intelligent level of monitoring. The camera uses the running algorithm model to process the video data it collects and receives, generating detection results and enabling intelligent analysis and processing of the surveillance video. For example, it can detect target objects such as people, vehicles, or flames in the video in real time and issue timely alarms.

[0020] This invention provides data security and privacy protection: It encrypts transmitted video data to ensure data security during transmission, effectively preventing data leakage and malicious tampering. Data is transmitted and processed within the local area network, avoiding the privacy risks associated with uploading sensitive data to the cloud and complying with data protection regulations.

[0021] The present invention optimizes the system and reduces costs: each camera only runs one algorithm model, avoiding repeated calculations and improving computing power utilization. Compared with the traditional single-camera single algorithm or multi-camera plus end-side edge computing gateway calculation method, the present invention can utilize computing power resources more efficiently. It reduces dependence on cloud computing power and reduces cloud costs. For large-scale monitoring systems, this cost advantage is particularly obvious and can greatly save operating costs. The end-side edge computing gateway device is eliminated, reducing the hardware cost of the system. Compared with the multi-camera plus end-side edge computing gateway calculation method, the present invention does not require the additional purchase and maintenance of edge computing gateway devices, simplifying the system architecture.

[0022] Improved real-time performance and reliability: Video data is transmitted via the RTSP protocol, ensuring real-time video data transmission, allowing the monitoring system to respond promptly to changes in the monitoring scene.

[0023] Use TCP / IP or UDP communication protocols for data transmission. Select the appropriate protocol based on actual needs to balance the reliability and real-time performance of data transmission. For example, TCP / IP can be selected in scenarios with high reliability requirements, while UDP can be selected in scenarios with high real-time requirements.

[0024] Enhanced flexibility and scalability: The cloud platform dynamically allocates algorithm models based on user needs, allowing cameras to flexibly run different algorithm models, improving system flexibility. This facilitates future upgrades and expansions of the surveillance system. New cameras can be easily added or algorithm models updated to adapt to changing surveillance needs, improving system scalability. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a schematic diagram of the data interaction process between the camera of the present invention, the cloud platform and the business end. DETAILED DESCRIPTION

[0026] The present invention is further described below with reference to the accompanying drawings and specific embodiments.

[0027] Reference Figure 1 The present invention provides a method for sharing surveillance videos, which is applied to a surveillance system composed of multiple cameras and a cloud platform, wherein the multiple cameras are connected via a local area network (LAN). The method comprises the following steps: Step S1: Camera detection and information reporting: After multiple cameras are connected to the same local area network, they detect the existence of other cameras (for example, by device ID or IP address identification) through the local area network communication module (such as a broadcast mechanism based on the TCP / IP protocol), and report the number of detected cameras, device identification and other information to the cloud platform.

[0028] Step S2: The cloud platform dynamically assigns an algorithm model: The cloud platform assigns a specific algorithm model to each camera based on the scenario selected by the user (such as "shopping mall security" or "campus abnormal warning"). For example: If the user selects "Shopping Mall Security" and needs to detect human figures, flames, and fighting behaviors, the cloud platform will assign the "Human Figure Detection Model" to Camera 1, the "Flame Detection Model" to Camera 2, and the "Fighting Recognition Model" to Camera 3.

[0029] Step S3: Cameras run the algorithm model: Each camera's NPU (neural processing unit) loads and runs the algorithm model assigned by the cloud platform. Since the camera's NPU computing power is limited to 0.5TOPS (supporting only a single algorithm), distributed allocation achieves multi-dimensional detection coverage.

[0030] Step S4: Sharing Video Data Within the LAN. Each camera captures video data and divides it into two streams: a main stream (1920 x 1080, for local storage or high-definition viewing) and a substream (640 x 360, for low-bandwidth transmission). The camera transmits the encrypted substream data to other cameras via the RTSP protocol (for example, camera 1 sends the substream to cameras 2 and 3, while simultaneously receiving substreams from cameras 2 and 3). The transmission process uses the AES encryption algorithm to ensure data security, and supports TCP / IP or UDP protocols (TCP for reliable transmission and UDP for low-latency scenarios).

[0031] Step S5: Local data processing and result generation: After receiving the encrypted sub-stream from other devices, the camera decrypts it and stores it in a circular buffer (temporarily storing data to avoid overflow), extracts frames at a fixed frequency (e.g., 10 frames per second), and processes the extracted frame images using its own algorithm model (for example, camera 1 uses the human detection model to process its own sub-stream and the sub-streams of cameras 2 and 3 to identify human targets within the coverage area of ​​all cameras), generating a detection result containing "camera ID, timestamp, target coordinates, and confidence level."

[0032] Step S6: Result upload and distribution: The camera encapsulates the detection results in JSON format and sends them to the cloud platform. The cloud platform distributes the results to the corresponding users (such as the mall security app and the campus duty room client) based on the camera ID in the results.

[0033] The monitoring system of the present invention is composed of multiple cameras and a cloud platform, and the multiple cameras are interconnected through a local area network.

[0034] Each camera includes Algorithm execution module: Executes the algorithm model assigned by the cloud platform, enabling the camera to possess specific detection capabilities. Video transmission module: Sends and receives substream video data via the RTSP protocol, enabling video data sharing between cameras. Data encryption module: Encrypts and decrypts transmitted video data to ensure data security. Result packaging module: Packages detection results into JSON format for easy parsing and processing by the cloud platform.

[0035] NPU (Neural Network Processing Unit): Each camera is equipped with an NPU, and each NPU only runs a single algorithm model, improving the operating efficiency of the algorithm model.

[0036] The cloud platform is used to receive information about other cameras in the local area network reported by the camera, assign an algorithm model to the camera according to the scene selected by the user, receive detection results and distribute them to the corresponding users.

[0037] The cloud platform also includes a decision-making module for performing integrated analysis and decision-making on the detection results and distributing them to different business terminals (such as APP or client), such as business 1, business 2, business 3, etc.

[0038] The present invention also provides an electronic device comprising a processor and a memory, wherein the memory stores a computer program, and wherein the processor, when executing the computer program, implements the steps of the aforementioned surveillance video sharing method. For example, the processor and memory in a camera can constitute an electronic device that executes the surveillance video sharing method to implement functions such as video data collection, processing, transmission, and detection.

[0039] The following is further described through specific examples: Example 1: Shopping mall security Scenario: Deploy three cameras in a large shopping mall (the specific number depends on the actual situation): entrance (A), parking lot (B), and food court (C). Users can select human detection, vehicle detection, and flame detection.

[0040] Implementation process: Camera detection: Camera A, Camera B, Camera C (corresponding Figure 1 Camera 1, Camera 2, and Camera 3 in the network detect each other through UPnP and report to the cloud platform.

[0041] Algorithm allocation: The cloud platform assigns human detection to camera A, vehicle detection to camera B, and flame detection to camera C.

[0042] Video sharing: Each camera generates a sub-stream (640*360), which is shared via RTSP after encryption and stored in a ring buffer after decryption.

[0043] Data processing: Camera A processes the A, B, and C sub-streams to detect human figures, camera B detects vehicles, and camera C detects flames.

[0044] Result upload: The results are sent to the cloud platform in JSON format; Result distribution: If the cloud platform detects a flame, it triggers the fire protection system and notifies the security app; if a suspicious person is detected, an alarm is sent.

[0045] Effect: A 3×3 detection matrix is ​​formed, the false alarm rate is reduced by about 40%, and the detection delay is <100ms.

[0046] Example 2: Bank vault security scenario The system consists of four cameras (D, E, F, and G), a cloud platform, and a bank security system. The cameras are deployed at the vault entrance, inside, in the corridors, and in the monitoring room, supporting facial recognition, left-behind object detection, and abnormal motion analysis algorithms.

[0047] The cloud platform assigns a "face recognition model" to camera D (entrance), a "left-behind object detection model" to camera E (interior), an "abnormal motion analysis model" to camera F (corridor), and a "backup algorithm model" to camera G (monitoring room, dynamically switchable). Substream transmission uses the UDP protocol (to reduce latency), and the encryption method is upgraded to RSA+AES hybrid encryption (for high security requirements). Camera D processes all substreams using the face recognition model, identifying an "unauthorized person at the entrance," and camera F detects a "person running in the corridor." The cloud platform's decision-making module triggers a "vault intrusion warning," distributes the results to the bank's security terminal, and activates the access control system to lock the entrance.

[0048] Example 3: Residential Area Monitoring Scenario: A residential area deploys cameras at the perimeter (A) and public areas (B and C). The user selects intrusion detection, fire detection, and pet detection.

[0049] Implementation process: Camera detection: Cameras A, B, and C detect each other and report to the cloud platform.

[0050] Algorithm assignment: Camera A runs intrusion detection, Camera B runs fire detection, and Camera C runs pet detection.

[0051] Video sharing: Each camera shares the encrypted sub-stream, which is then decrypted and stored in a ring buffer.

[0052] Data processing: Camera A detects abnormal movement within the perimeter, Camera B detects smoke, and Camera C identifies pets.

[0053] Result upload and distribution: The results are sent to the cloud platform and distributed to the residents' APP.

[0054] Effect: If an intrusion is detected, residents receive a real-time alert; if a fire is detected, a fire alarm is triggered; if a pet is identified, residents are notified.

[0055] Example 4: Industrial Facility Monitoring Scenario: An industrial facility deploys cameras on production lines (A), warehouses (B), and perimeters (C). Users select protective equipment (PPE) detection, object detection, and intrusion detection.

[0056] Implementation process: Camera detection: Camera A, Camera B, and Camera C detect cameras in the local area network and report to the cloud platform.

[0057] Algorithm assignment: Camera A runs PPE detection, Camera B runs object detection, and Camera C runs intrusion detection.

[0058] Video sharing: Share encrypted sub-stream and process it after decryption.

[0059] Data processing: Camera A detects whether workers are wearing safety gear, Camera B monitors inventory status, and Camera C identifies perimeter anomalies.

[0060] Result upload: The results are sent to the cloud platform and distributed to the management terminal.

[0061] Effect: If a worker isn't wearing a hard hat, an alarm is triggered; if inventory is abnormal, management is notified; if an intrusion is detected, security measures are initiated.

[0062] Example 5: Urban Traffic Management Scenario: Cameras are deployed at urban intersections, highways, and parking lots, with users selecting traffic flow analysis, accident detection, and parking space monitoring.

[0063] Implementation process: Camera detection: The camera detects other devices in the local area network and reports to the cloud platform.

[0064] Algorithm assignment: Intersection cameras run vehicle counting, highway cameras run accident detection, and parking lot cameras run parking space detection.

[0065] Video sharing: Share sub-stream and process after decryption.

[0066] Data processing: Intersection cameras count vehicles, highway cameras detect accidents, and parking lot cameras identify vacant spaces.

[0067] Result upload: The results are sent to the cloud platform to optimize traffic signals or notify traffic police.

[0068] Effect: If an accident is detected, adjust the signal lights and notify the traffic police; if an empty space is identified, update the navigation system.

[0069] Example 6: Home Security Scenario: A home deploys cameras at the front door (A) and backyard (B), and the user selects facial recognition and package detection.

[0070] Implementation process: Camera detection: Camera A and Camera B detect each other and report to the cloud platform.

[0071] Algorithm assignment: Camera A runs face recognition, and Camera B runs package detection.

[0072] Video sharing: Share encrypted sub-stream and process it after decryption.

[0073] Data processing: Camera A identifies family members or strangers, and Camera B detects packages.

[0074] Result upload: The results are sent to the cloud platform and distributed to the user's mobile phone.

[0075] Effect: If a stranger is detected, an alarm is sent; if a package is detected, the user is notified to pick it up.

[0076] As can be seen from the above specific implementation scheme, the present invention reduces hardware costs: there is no need to deploy an edge computing gateway, and equipment investment can be saved through distributed collaboration of cameras; This invention improves computing power utilization: each camera runs a single algorithm, and through data sharing, an "N×N" detection matrix is ​​formed (N is the number of cameras), avoiding repeated calculations; This invention reduces cloud load: by only uploading detection results (instead of original videos), the cloud computing power required for 100 cameras is reduced to 10% of the traditional solution; This invention optimizes privacy and real-time performance: raw video data is transmitted within the local area network, and sensitive information (such as faces) does not leave the local area network, complying with regulations such as GDPR; detection delay is <100ms, meeting real-time warning needs.

[0077] This invention enables efficient sharing and collaborative processing: it enables efficient video data sharing between multiple cameras, transmitting encrypted substream video data via a local area network, ensuring efficient data transmission while reducing network bandwidth requirements. Video information can be shared between cameras in real time, avoiding the problem of information silos. The cameras collaborate to run different algorithm models, forming an integrated monitoring network. For example, one camera can run a human detection algorithm while another runs a vehicle detection algorithm. These algorithms can work together to achieve comprehensive monitoring of complex scenes.

[0078] This invention provides intelligent monitoring: the cloud platform assigns corresponding algorithm models to cameras based on user-selected scenarios, enabling the monitoring system to flexibly adapt to diverse monitoring needs. Users can select algorithm models for human detection, vehicle detection, or flame detection based on the specific scenario, enhancing the intelligent level of monitoring. The camera uses the running algorithm model to process the video data it collects and receives, generating detection results and enabling intelligent analysis and processing of the surveillance video. For example, it can detect target objects such as people, vehicles, or flames in the video in real time and issue timely alarms.

[0079] This invention provides data security and privacy protection: It encrypts transmitted video data to ensure data security during transmission, effectively preventing data leakage and malicious tampering. Data is transmitted and processed within the local area network, avoiding the privacy risks associated with uploading sensitive data to the cloud and complying with data protection regulations.

[0080] The present invention optimizes the system and reduces costs: each camera only runs one algorithm model, avoiding repeated calculations and improving computing power utilization. Compared with the traditional single-camera single algorithm or multi-camera plus end-side edge computing gateway calculation method, the present invention can utilize computing power resources more efficiently. It reduces dependence on cloud computing power and reduces cloud costs. For large-scale monitoring systems, this cost advantage is particularly obvious and can greatly save operating costs. The end-side edge computing gateway device is eliminated, reducing the hardware cost of the system. Compared with the multi-camera plus end-side edge computing gateway calculation method, the present invention does not require the additional purchase and maintenance of edge computing gateway devices, simplifying the system architecture.

[0081] Improved real-time performance and reliability: Video data is transmitted via the RTSP protocol, ensuring real-time video data transmission, allowing the monitoring system to respond promptly to changes in the monitoring scene.

[0082] Use TCP / IP or UDP communication protocols for data transmission. Select the appropriate protocol based on actual needs to balance the reliability and real-time performance of data transmission. For example, TCP / IP can be selected in scenarios with high reliability requirements, while UDP can be selected in scenarios with high real-time requirements.

[0083] Enhanced flexibility and scalability: The cloud platform dynamically allocates algorithm models based on user needs, allowing cameras to flexibly run different algorithm models, improving system flexibility. This facilitates future upgrades and expansions of the surveillance system. New cameras can be easily added or algorithm models updated to adapt to changing surveillance needs, improving system scalability.

[0084] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made by using the contents of the present invention description and drawings under the inventive concept of the present invention, or direct / indirect application in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A method for sharing surveillance videos, applied to a surveillance system consisting of multiple cameras and a cloud platform, wherein the multiple cameras are connected via a local area network, characterized in that: The method comprises the following steps: Step S1: Each of the multiple cameras detects other cameras in the local area network and reports the detected information of the other cameras to the cloud platform; Step S2: The cloud platform assigns a corresponding algorithm model to each camera based on the scene selected by the user, where each algorithm model is configured to detect a specific target object or scene; Step S3: Each camera runs its assigned algorithm model; Step S4: Each camera sends the video data collected by itself to other cameras through the local area network, and receives the video data sent by other cameras at the same time; Step S5: The camera uses the running algorithm model to process the video data collected and received by itself to generate a detection result; Step S6: Each camera sends the detection results to the cloud platform.

2. The method for sharing surveillance video according to claim 1, characterized in that: In step S2 , the scenario selected by the user includes at least one of human detection, vehicle detection, or flame detection, which is used to determine the type of algorithm model assigned to the camera.

3. The method for sharing surveillance video according to claim 1, characterized in that: The video data is sub-stream data with a lower resolution than the main stream, wherein each camera is divided into two streams, wherein the main stream resolution is 1920*1080 and the sub-stream resolution is 640*360.

4. The method for sharing surveillance video according to claim 3, characterized in that: In step S4, the resolution video data of each sub-stream is encrypted and transmitted to other cameras through the RTSP protocol at the same time. After receiving the data through RTSP, each camera decrypts it and stores it in a ring buffer. After extracting frames to detect the video data, the output detection results are encapsulated in JSON format and sent to the cloud platform. The cloud platform then distributes the detection results to the corresponding users based on the camera ID of the detection results.

5. The method for sharing surveillance video according to claim 4, characterized in that: The detection result includes: camera identification, timestamp, target coordinates and confidence level.

6. A monitoring system, characterized in that: include: Multiple cameras and cloud platforms; The multiple cameras are interconnected via a local area network, and each camera includes: an algorithm running module for executing the algorithm model assigned by the cloud platform; a video transmission module for sending and receiving sub-stream video data; a data encryption module for encrypting and decrypting transmitted data; and a result encapsulation module for encapsulating the detection results into JSON format data; The cloud platform is used to receive information about other cameras in the local area network reported by the camera, assign an algorithm model to the camera according to the scene selected by the user, receive detection results and distribute them to the corresponding users.

7. The monitoring system according to claim 6, characterized in that: The camera is configured with an NPU, and each NPU only runs a single algorithm model.

8. The monitoring system according to claim 6, characterized in that: The video transmission module supports TCP / IP or UDP protocol.

9. The monitoring system according to claim 7, characterized in that: The cloud platform also includes a decision-making module for performing fusion analysis on the detection results and distributing them to different business terminals after making decisions.

10. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and when the processor executes the computer program, the method implements the steps of any one of claims 1 to 5.

Images