Remote monitoring method, device, equipment and storage medium of camera

Abstract

The present invention relates to the technical field of data processing, and discloses a remote monitoring method, device, equipment and storage medium for a camera. The remote monitoring method for the camera comprises: collecting video images of multiple monitoring scenes through a preset composite monitoring network, and encoding and processing each of the video images respectively to obtain a corresponding video stream file; classifying the video stream file to obtain a first data segment and a second data segment; encrypting the first data segment and the second data segment respectively to obtain an encrypted first data segment and an encrypted second data segment; the present invention uses a composite monitoring network to collect video images, has good scalability, is easy to add new monitoring points and smart cameras, achieves a wider range of monitoring coverage, and adapts to changing monitoring needs.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a remote monitoring method, device, equipment and storage medium for a camera. Background Art

[0002] In modern society, with the enhancement of public safety awareness and the development of technology, remote monitoring systems have been widely used in various fields. Especially in public areas such as shopping malls, transportation, and residential areas, by installing surveillance cameras, various events can be monitored in real time, providing a basis for the prevention and handling of accidents. Traditional remote monitoring systems mainly rely on surveillance cameras connected by physical lines and local servers to store and process recorded video data. While these systems realize the basic functions of monitoring, they also expose some technical and application limitations.

[0003] In the prior art, some cloud computing-based remote monitoring solutions have begun to be developed and used. These systems upload monitoring data to cloud servers and use the powerful computing and storage capabilities of cloud computing to process and analyze video data, which can more effectively manage and utilize monitoring resources. However, these systems still have shortcomings in terms of secure transmission and efficient processing of video data. In particular, in terms of data encryption, data security during transmission, and how to efficiently classify and analyze video data according to the content of the monitoring scene, the existing technical solutions often lack effective solutions.

[0004] The main technical defects of the existing technology are that the efficiency and security of its video data processing and transmission are insufficient. First, when uploading video data to the cloud server, the traditional remote monitoring system often adopts simple encryption methods, which makes it difficult to ensure the security of the data during the transmission process. Second, the existing cloud-based monitoring solution lacks an effective data classification mechanism when processing a large amount of monitoring video data, resulting in the inability to quickly locate and analyze emergencies, affecting the monitoring system's response time to emergencies. Third, the existing solutions rely more on regularized pattern recognition algorithms in the analysis and processing of video data, lack the ability to deeply learn and analyze complex scenes, and limit the effectiveness of the monitoring system in identifying new abnormal behaviors. Therefore, there is a need for a remote monitoring method that can improve the efficiency of secure transmission of video data, effectively classify and process monitoring video data, and accurately analyze abnormal events. Summary of the invention

[0005] The present invention provides a remote monitoring method, device, equipment and storage medium for a camera, which are used to solve the above-mentioned technical problems.

[0006] A first aspect of the present invention provides a remote monitoring method for a camera, the remote monitoring method for a camera comprising:

[0007] The video images of multiple monitoring scenes are collected through a preset composite monitoring network, and each of the video images is encoded and processed to obtain a corresponding video stream file; wherein the composite monitoring network is composed of a plurality of intelligent monitoring cameras connected;

[0008] The video stream file is classified to obtain a first data segment and a second data segment; wherein the first data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where an emergency occurs; and the second data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where no emergency occurs;

[0009] Encrypting the first data segment and the second data segment respectively to obtain an encrypted first data segment and an encrypted second data segment, randomly generating a plurality of different obfuscated data segments, and combining the plurality of different obfuscated data segments, the encrypted first encrypted data, and the encrypted second encrypted data according to a preset arrangement rule to obtain an encrypted video stream file, and uploading the encrypted video stream file to a cloud server;

[0010] The cloud server decrypts the encrypted video stream file to obtain a decrypted video stream file; wherein the cloud server pre-stores a decryption algorithm that is the inverse of the encryption algorithm used to encrypt the first data segment and the second data segment;

[0011] The decrypted video stream files are detected based on the deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report.

[0012] Optionally, in a first implementation of the first aspect of the present invention, the intelligent surveillance camera includes a remote image sensor, a video encoder, and an adaptive video stream dynamic adjustment module;

[0013] The video images of multiple monitoring scenes are collected through a preset composite monitoring network, and each of the video images is encoded and processed to obtain a corresponding video stream file, including:

[0014] Configuring each intelligent monitoring camera in the composite monitoring network to obtain a plurality of configured intelligent monitoring cameras;

[0015] The remote image sensor is used to collect video images in multiple monitoring scenes in real time, and the collected video images are converted into digital signals by the remote image sensor to obtain a video data stream;

[0016] Encoding the video data stream according to a preset video encoding standard by a video encoder to obtain an encoded video data stream;

[0017] The encoded video data stream is dynamically adjusted through the adaptive video stream dynamic adjustment module to obtain the corresponding video stream file.

[0018] Optionally, in a second implementation of the first aspect of the present invention, encrypting the first data segment and the second data segment respectively to obtain the encrypted first data segment and the encrypted second data segment includes:

[0019] Based on a preset data sensitivity assessment rule, a sensitivity assessment is performed on the first data segment to obtain a first encryption requirement for the first data segment.

[0020] Based on a preset compliance assessment rule, a compliance assessment is performed on the second data segment to obtain a second encryption requirement of the second data segment;

[0021] Based on the first encryption requirement of the first data segment, selecting a first encryption algorithm matching the first encryption requirement from a preset encryption algorithm database, encrypting the first data segment by using the first encryption algorithm to obtain an encrypted first data segment;

[0022] Based on the second encryption requirement of the second data segment, a second encryption algorithm matching the second encryption requirement is selected from a preset encryption algorithm database, and the second data segment is encrypted by the second encryption algorithm to obtain an encrypted second data segment.

[0023] Optionally, in a third implementation of the first aspect of the present invention, the deep learning model deployed on the cloud server detects the decrypted video stream file to obtain an abnormal monitoring scene analysis report, including:

[0024] Input the decrypted video stream file into a deep learning model; wherein the deep learning model at least includes an input layer, a segmentation layer, a convolution layer, an activation layer, a pooling layer, a fully connected layer, and an output layer;

[0025] The decrypted video stream file is normalized through the input layer to obtain a normalized video stream;

[0026] The normalized video stream is segmented by the segmentation layer to obtain a plurality of static images;

[0027] Extracting features from each static image using a plurality of filters on the convolutional layer to obtain a first image feature set;

[0028] Learning the image feature set through a nonlinear activation function on the activation layer to obtain a second image feature set;

[0029] Performing maximum pooling on each second image feature region in the second image feature set through the pooling layer to obtain a third image feature set;

[0030] Flatten the third image feature set through the fully connected layer to obtain classified feature representations;

[0031] The softmax function on the output layer makes a decision output on each feature representation after classification to obtain an abnormal monitoring scene analysis report; wherein the abnormal monitoring scene analysis report at least includes a probability score of an abnormal monitoring scene appearing in the video stream at each moment.

[0032] Optionally, in a fourth implementation method of the first aspect of the present invention, the configuration of each smart surveillance camera in the composite monitoring network includes at least assigning an IP address to each smart camera, registering each smart camera to the management platform of the monitoring center, and setting the monitoring parameters of each smart camera.

[0033] A second aspect of the present invention provides a remote monitoring device for a camera, the remote monitoring device for a camera comprising:

[0034] The acquisition module is used to acquire video images of multiple monitoring scenes through a preset composite monitoring network, and encode each of the video images to obtain a corresponding video stream file; wherein the composite monitoring network is composed of a plurality of intelligent monitoring cameras connected;

[0035] A classification module, used to classify the video stream file to obtain a first data segment and a second data segment; wherein the first data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where an emergency occurs; and the second data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where no emergency occurs;

[0036] An encryption module is used to encrypt the first data segment and the second data segment respectively to obtain the encrypted first data segment and the encrypted second data segment, randomly generate a plurality of different obfuscated data segments, and combine the plurality of different obfuscated data segments, the encrypted first encrypted data, and the encrypted second encrypted data according to a preset arrangement rule to obtain an encrypted video stream file, and upload the encrypted video stream file to a cloud server;

[0037] A decryption module, used for the cloud server to decrypt the encrypted video stream file to obtain a decrypted video stream file; wherein the cloud server pre-stores a decryption algorithm that is the inverse of the encryption algorithm used to encrypt the first data segment and the second data segment;

[0038] The detection module is used to detect the decrypted video stream files based on the deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report.

[0039] The third aspect of the present invention provides a remote monitoring device for a camera, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the remote monitoring device for the camera executes the above-mentioned remote monitoring method for the camera.

[0040] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned remote monitoring method of a camera.

[0041] The technical solution provided by the present invention has the following beneficial effects: the present invention provides a remote monitoring method, device, equipment and storage medium of a camera, collects video images of multiple monitoring scenes through a preset composite monitoring network, and encodes each of the video images to obtain a corresponding video stream file; classifies the video stream file to obtain a first data segment and a second data segment; encrypts the first data segment and the second data segment respectively to obtain an encrypted first data segment and an encrypted second data segment, randomly generates a plurality of different obfuscated data segments, and combines the plurality of different obfuscated data segments, the encrypted first encrypted data and the encrypted second encrypted data according to a preset arrangement rule to obtain an encrypted video stream file, and uploads the encrypted video stream file to a cloud server; the cloud server decrypts the encrypted video stream file to obtain a decrypted video stream file; detects the decrypted video stream file based on a deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report. The present invention enhances the security of video data in the process of uploading to the cloud server by encrypting the video stream data and combining the obfuscated data segments, and effectively prevents data leakage and tampering. By classifying different data segments, the monitoring data of emergencies and normal monitoring data can be processed more specifically. In particular, by giving priority to data segments where emergencies occur, emergency situations can be acquired and responded to more quickly, improving the emergency response capabilities of the monitoring system. Data decryption and deep learning model analysis are performed by the cloud server, so that data processing is not limited by local hardware and the high-performance computing resources of cloud computing are fully utilized. The deep learning model is used to detect and analyze the decrypted video stream files, which can realize efficient and accurate abnormal behavior recognition for complex monitoring scenarios, and generate abnormal monitoring scene analysis reports in a timely manner, thereby improving the intelligence level of the monitoring system. The present invention utilizes a composite monitoring network to collect video images, has good scalability, and is convenient for adding new monitoring points and smart cameras, so as to achieve a wider range of monitoring coverage and adapt to changing monitoring needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A schematic diagram of an embodiment of a remote monitoring method of a camera in an embodiment of the present invention;

[0043] Figure 2 Schematic diagram of a remote monitoring device for a camera in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] Embodiments of the present invention provide a remote monitoring method, device, equipment and storage medium for a camera. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0045] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the remote monitoring method of a camera in an embodiment of the present invention includes:

[0046] Step 101: Collect video images of multiple monitoring scenes through a preset composite monitoring network, and encode each of the video images to obtain a corresponding video stream file; wherein the composite monitoring network is composed of a plurality of intelligent monitoring cameras connected;

[0047] It is understandable that the execution subject of the present invention may be a remote monitoring device of a camera, or may be a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0048] Specifically, in step 101, the process of collecting video images through the preset composite monitoring network can be expanded and refined into the following sub-steps:

[0049] Deployment of smart surveillance cameras:

[0050] Multiple smart surveillance cameras are installed in different surveillance scenarios. These cameras can capture real-time images within their respective fields of view, and usually choose appropriate camera positions and angles according to surveillance needs.

[0051] Configure network connections:

[0052] Connect smart surveillance cameras to a composite surveillance network. This usually involves network configuration, ensuring that the cameras are stably connected to the central surveillance system via a wired or wireless network.

[0053] Video capture:

[0054] The camera continuously captures video images according to preset configurations (such as resolution, frame rate, etc.). Smart surveillance cameras can have functions such as automatic focus and light adjustment to ensure image quality.

[0055] Real-time video encoding:

[0056] The camera or connected monitoring system will encode the captured video in real time to optimize storage and transmission efficiency. Common encoding formats include H.264, H.265, etc.

[0057] Generate video stream file:

[0058] The encoded video images are converted into video stream files and can be stored in a local storage device or transmitted to a remote server via a network as needed.

[0059] Step 102: Classify the video stream file to obtain a first data segment and a second data segment; wherein the first data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where an emergency occurs; and the second data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where no emergency occurs;

[0060] Specifically, the execution of step 102 is intended to distinguish the portion of the video stream file containing the emergency (the first data segment) from the portion of the regular monitoring content (the second data segment). This step can be expanded and refined through the following sub-steps:

[0061] Expand and refine steps

[0062] Video data reception and buffering:

[0063] The monitoring system first receives the video stream files transmitted by all smart cameras and buffers them on the server.

[0064] Event detection algorithm deployment:

[0065] Deploy event detection algorithms, which can be rule-based systems or machine learning / deep learning models, to automatically identify abnormal behaviors or other defined emergencies in videos.

[0066] Real-time video analysis:

[0067] Video data is analyzed in real time, with algorithms identifying specific patterns, motion, or other defined abnormal indicators in the footage that may indicate the occurrence of an emergency.

[0068] Data segment classification:

[0069] According to the result of event detection, the algorithm automatically divides the video stream into a first data segment containing the emergency event and a second data segment of the regular monitoring screen.

[0070] Data storage and annotation:

[0071] The two types of video data can be stored in different directories or databases, and the data containing emergencies can be marked to facilitate rapid retrieval and analysis in the future.

[0072] Example

[0073] Taking the mall surveillance system as an example, the mall hopes to quickly identify and respond to emergencies, such as a child getting lost, a customer falling, or an emergency evacuation:

[0074] Receiving video stream:

[0075] The security server continuously receives video stream data from surveillance cameras in all corners of the mall and caches the data in temporary storage.

[0076] Deployment of intelligent detection:

[0077] A deep learning model is trained to detect abnormal behavior in videos, such as fast-moving people, objects that remain stationary for long periods of time, or obvious falls.

[0078] Analyze and categorize videos:

[0079] The video stream is fed into the model for analysis in real time. Whenever the detection algorithm identifies a sudden event, the video segment is classified as the first data segment.

[0080] Video data storage:

[0081] The detected first data segment is marked and saved in a specific folder, while the video stream without sudden events is classified as the second data segment and saved according to the conventional procedure.

[0082] Step 103: Encrypt the first data segment and the second data segment respectively to obtain an encrypted first data segment and an encrypted second data segment, randomly generate a plurality of different obfuscated data segments, and combine the plurality of different obfuscated data segments, the encrypted first encrypted data, and the encrypted second encrypted data according to a preset arrangement rule to obtain an encrypted video stream file, and upload the encrypted video stream file to a cloud server;

[0083] Specifically, the execution of step 103 is mainly to protect the security and confidentiality of the video stream data and prevent unauthorized third party access during uploading to the cloud server or after storage. This step can be expanded and refined through the following sub-steps:

[0084] Select an encryption algorithm:

[0085] Select an algorithm suitable for video data encryption, such as a symmetric encryption algorithm such as AES (Advanced Encryption Standard) or an asymmetric encryption algorithm such as RSA.

[0086] Encrypt the first data segment and the second data segment:

[0087] The first data segment (containing the emergency event) and the second data segment (regular monitoring content) are encrypted using the selected encryption algorithm respectively.

[0088] Obfuscated data segment generation:

[0089] Multiple obfuscated data segments are randomly generated. These data segments have no actual content but are similar to real video data segments in structure.

[0090] Combined data segment:

[0091] The encrypted data segments are combined with the obfuscated data segments according to preset arrangement rules to create an encrypted video stream that is difficult to crack directly.

[0092] Arrangement of encrypted video stream files:

[0093] Integrate and package the arranged video stream files, maintain the integrity of the files and prepare them for uploading.

[0094] Step 104: the cloud server decrypts the encrypted video stream file to obtain a decrypted video stream file; wherein the cloud server pre-stores a decryption algorithm that is the inverse of the encryption algorithm used to encrypt the first data segment and the second data segment;

[0095] Specifically, step 104 involves a decryption process on the cloud server, which aims to restore the uploaded encrypted video stream file to the original video file for authorized users to view and analyze. The following is an expansion and detailed description of this step:

[0096] Verify the security of uploaded files:

[0097] After the server receives the encrypted video stream file, it first performs an integrity and security check on the file, such as verifying the file hash value to ensure that it has not been tampered with during transmission.

[0098] Locate and separate obfuscated data:

[0099] According to the pre-set arrangement rules, the cloud server will identify and separate the encrypted content and obfuscated data segments in the video stream file.

[0100] Get the decryption key:

[0101] The server accesses a secure key management system to obtain the decryption key corresponding to the encryption algorithm. This is usually achieved through a secure protocol, ensuring that the transmission and use of the key is also secure.

[0102] Decrypted video data:

[0103] The encrypted first data segment and the second data segment are decrypted using a pre-stored decryption algorithm and its corresponding key.

[0104] Reassemble video content:

[0105] The decrypted first and second data segments are reassembled in sequence into a complete video file, excluding all obfuscated data.

[0106] File Validation and Storage:

[0107] The decrypted video file is subjected to another security check and stored on the server in accordance with the format requirements for subsequent access and analysis.

[0108] Example

[0109] Taking a bank's ATM monitoring system as an example, the bank hopes to ensure the confidentiality of all surveillance videos while facilitating the rapid review of abnormal transaction events on the cloud platform:

[0110] Upload file security check:

[0111] The cloud platform receives the surveillance video stream file in front of the ATM and verifies it to ensure that the file has not been tampered with during the upload process.

[0112] Obfuscated data separation:

[0113] The server identifies the arrangement rules to distinguish which data belongs to the real surveillance video data and which are obfuscated data segments added for security.

[0114] Decryption key acquisition:

[0115] The cloud server securely obtains the key for decrypting the video data through the bank's key management system.

[0116] Perform video data decryption:

[0117] The cloud uses the obtained key to decrypt the video stream file through the AES decryption algorithm.

[0118] Video data reassembly:

[0119] After decryption, the original video segments are reassembled in the correct order, and the obfuscated redundant data is excluded.

[0120] Video storage and verification:

[0121] The video data that has been decrypted and integrity-verified is stored in the secure database of the bank's cloud platform. When an abnormal ATM transaction is detected, the relevant department can quickly pull the video at the relevant time point for review.

[0122] Step 105: Detect the decrypted video stream file based on the deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report.

[0123] Specifically, step 105 involves using a deep learning model on a cloud server to analyze the decrypted video stream file in order to identify and report abnormal behaviors or events in the surveillance scene. The following is an expansion and refinement of this step:

[0124] Prepare the deep learning model:

[0125] Deploy a trained deep learning model on a cloud server, which can be a model for a specific task (such as pedestrian detection, abnormal behavior recognition, etc.).

[0126] Video preprocessing:

[0127] Perform necessary preprocessing on the decrypted video stream file, such as adjusting the resolution, cropping the area, normalizing the data format, etc., to adapt to the input requirements of the deep learning model.

[0128] Video frame extraction:

[0129] The video stream is broken down into a sequence of frames, and the deep learning model analyzes these individual frames.

[0130] Run the deep learning model:

[0131] Apply deep learning models to analyze videos frame by frame to detect behavioral patterns, objects, or other specific features to identify potential anomalies.

[0132] Analysis results aggregation:

[0133] The single-frame analysis results of the model are aggregated to generate a comprehensive analysis report for the entire video segment.

[0134] Generate anomaly monitoring report:

[0135] If an anomaly is detected, the detailed analysis results are associated with the frame or video segment where the anomaly is located, and an abnormal monitoring scene analysis report is generated.

[0136] Another embodiment of the remote monitoring method of a camera in the embodiment of the present invention includes: the intelligent monitoring camera includes a remote image sensor, a video encoder, and an adaptive video stream dynamic adjustment module;

[0137] The video images of multiple monitoring scenes are collected through a preset composite monitoring network, and each of the video images is encoded and processed to obtain a corresponding video stream file, including:

[0138] Configuring each intelligent monitoring camera in the composite monitoring network to obtain a plurality of configured intelligent monitoring cameras;

[0139] The remote image sensor is used to collect video images in multiple monitoring scenes in real time, and the collected video images are converted into digital signals by the remote image sensor to obtain a video data stream;

[0140] Encoding the video data stream according to a preset video encoding standard by a video encoder to obtain an encoded video data stream;

[0141] The encoded video data stream is dynamically adjusted through the adaptive video stream dynamic adjustment module to obtain the corresponding video stream file.

[0142] Specifically, an embodiment of the present invention provides a remote monitoring method of an intelligent monitoring camera, wherein the intelligent monitoring camera includes a remote image sensor, a video encoder, and an adaptive video stream dynamic adjustment module; the specific implementation steps of this embodiment are decomposed as follows:

[0143] Configure smart surveillance cameras:

[0144] Each camera is connected to the composite monitoring network through its built-in network interface module. The initial configuration of the camera is carried out in the network management center, including the setting of network parameters, identification of image sensors, video encoding settings and the allocation of access control permissions.

[0145] Collect video images:

[0146] The high-resolution remote image sensor built into the smart surveillance camera collects video images of the surveillance scene in real time and converts the analog visual signal into a digital signal. The sensor usually has advanced features such as automatic focus, exposure adjustment, and low-light enhancement.

[0147] Video data stream encoding:

[0148] The collected video data stream is processed by the built-in video encoder of the camera. The encoder compresses and encodes the signal according to the preset video encoding standard (such as H.264, H.265) to reduce the data bandwidth and storage space occupied.

[0149] Adaptive video streaming dynamic adjustment:

[0150] The encoded video data stream is processed by an adaptive video stream dynamic adjustment module, which adjusts the video stream quality in real time according to the current network conditions and bandwidth availability.

[0151] This module uses machine learning algorithms and prediction models to dynamically adjust the bitrate and picture quality parameters of the video stream, such as resolution and frame rate, according to real-time network changes to ensure that the smoothness and continuity of video transmission can be maintained even in the event of network fluctuations.

[0152] Generate video stream file:

[0153] The adaptively adjusted video stream is encapsulated into a data file format suitable for network transmission (such as MPEG-DASH, HLS), which includes video data and related metadata information such as timestamps and synchronization identifiers.

[0154] After the implementation of this technical solution, the following beneficial effects can be derived:

[0155] High-quality video surveillance: By deploying high-performance remote image sensors, smart surveillance cameras can capture high-resolution video images in real time, providing clearer image quality, making monitoring more accurate and facilitating event identification and analysis.

[0156] Optimization of coding efficiency: The use of advanced video encoders and coding standards, such as H.264 or H.265, can significantly compress video data streams, reduce the required storage space and network bandwidth, while maintaining video quality, and improve the cost-effectiveness of video surveillance systems.

[0157] Adaptive bandwidth management: The adaptive video stream dynamic adjustment module allows the encoded video data stream to adjust its transmission parameters in real time according to network conditions. For example, it reduces the bit rate of the video stream when the network is congested to ensure the continuity of the video stream, enhancing the stability and reliability of video surveillance transmission.

[0158] Flexibility of monitoring network: By uniformly configuring the intelligent monitoring cameras in the composite monitoring network, the monitoring system is highly flexible and scalable, making the monitoring network easy to expand and maintain, while simplifying the management of the monitoring system.

[0159] Another embodiment of the remote monitoring method of the camera in the embodiment of the present invention includes: encrypting the first data segment and the second data segment respectively to obtain the encrypted first data segment and the encrypted second data segment, including:

[0160] Based on a preset data sensitivity assessment rule, a sensitivity assessment is performed on the first data segment to obtain a first encryption requirement for the first data segment.

[0161] Based on a preset compliance assessment rule, a compliance assessment is performed on the second data segment to obtain a second encryption requirement of the second data segment;

[0162] Based on the first encryption requirement of the first data segment, selecting a first encryption algorithm matching the first encryption requirement from a preset encryption algorithm database, encrypting the first data segment by using the first encryption algorithm to obtain an encrypted first data segment;

[0163] Based on the second encryption requirement of the second data segment, a second encryption algorithm matching the second encryption requirement is selected from a preset encryption algorithm database, and the second data segment is encrypted by the second encryption algorithm to obtain an encrypted second data segment.

[0164] Specific, important term explanation:

[0165] Data sensitivity assessment rules: A set of predefined criteria used to determine the degree of sensitive information contained in the data, which may include personally identifiable information, financial information, and other data with varying levels of sensitivity.

[0166] First encryption requirement: The encryption requirement for the first data segment derived from the data sensitivity assessment rules, which may include the encryption strength, the encryption technology used, etc.

[0167] Compliance assessment rules: Rules set according to regulations, industry standards and organizational policies to assess the compliance of data and ensure that data processing complies with relevant laws, regulations and standards.

[0168] Second encryption requirement: The encryption requirement for the second data segment based on the compliance assessment rules, which may also specify the specific encryption method, encryption strength, etc.

[0169] Encryption algorithm database: A database containing various encryption algorithms, which can be a collection of different encryption technologies, such as symmetric encryption, asymmetric encryption, hash functions, etc.

[0170] Application scenarios:

[0171] This technical solution is suitable for scenarios where different levels of security protection are required for various surveillance video data segments. It is particularly suitable for financial institutions, government departments, security-sensitive enterprises, etc. In these scenarios, different video segments choose different encryption strategies based on the sensitivity of their content and compliance requirements.

[0172] Furthermore, for a monitoring center of a financial institution, when implementing the technical solution, the steps are as follows:

[0173] Evaluate the sensitivity of the first data segment:

[0174] The first data segment is scanned by automated tools to determine whether it contains highly sensitive content such as ATM operation interface, transaction information, etc., and the high encryption requirement is assessed accordingly.

[0175] Evaluate compliance for the second data segment:

[0176] Compliance checking software is used to analyze the second data segment to determine whether there is content in the surveillance footage that complies with general privacy and data protection regulations, such as pedestrians' faces, and to determine the need for moderate encryption.

[0177] Select the appropriate encryption algorithm:

[0178] According to the high-sensitivity encryption requirement of the first data segment, a strong encryption algorithm, such as AES-256, is selected from the encryption algorithm database.

[0179] For the medium compliance encryption requirements of the second data segment, a lighter encryption algorithm, such as AES-128, is selected.

[0180] Perform encryption operations:

[0181] AES-256 encryption is implemented on the first data segment to ensure high-strength encryption protection of sensitive information.

[0182] AES-128 encryption is implemented on the second data segment to ensure compliance while considering the reasonable allocation of computing resources.

[0183] After the implementation of this technical solution, the following beneficial effects can be derived:

[0184] Strengthen data security: Select encryption algorithms of different strengths based on data sensitivity and compliance requirements to ensure that sensitive information is strictly protected.

[0185] Save computing resources: By selecting appropriate encryption algorithms for different data segments, you can save computing resources and optimize the efficiency of encryption operations while ensuring security.

[0186] Improved data manageability: Systematic data assessment and encryption processes improve the controllability and transparency of data management at different levels of sensitivity.

[0187] Meet compliance requirements: Ensure that video surveillance data processing complies with regulatory requirements and avoid legal and compliance risks caused by data security and privacy issues.

[0188] Dynamic encryption policy adaptability: Based on the dynamic evaluation results of the data, the encryption policy can be adjusted according to the real-time situation, which improves the flexibility and responsiveness of security protection.

[0189] Another embodiment of the remote monitoring method of the camera in the embodiment of the present invention includes: the deep learning model deployed on the cloud server detects the decrypted video stream file to obtain an abnormal monitoring scene analysis report, including:

[0190] Input the decrypted video stream file into a deep learning model; wherein the deep learning model at least includes an input layer, a segmentation layer, a convolution layer, an activation layer, a pooling layer, a fully connected layer, and an output layer;

[0191] The decrypted video stream file is normalized through the input layer to obtain a normalized video stream;

[0192] The normalized video stream is segmented by the segmentation layer to obtain a plurality of static images;

[0193] Extracting features from each static image using a plurality of filters on the convolutional layer to obtain a first image feature set;

[0194] Learning the image feature set through a nonlinear activation function on the activation layer to obtain a second image feature set;

[0195] Performing maximum pooling on each second image feature region in the second image feature set through the pooling layer to obtain a third image feature set;

[0196] Flatten the third image feature set through the fully connected layer to obtain classified feature representations;

[0197] The softmax function on the output layer makes a decision output on each feature representation after classification to obtain an abnormal monitoring scene analysis report; wherein the abnormal monitoring scene analysis report at least includes a probability score of an abnormal monitoring scene appearing in the video stream at each moment.

[0198] Specific, important term explanation:

[0199] Normalization: scaling the data so that it falls into a small specific range. This process is commonly used in machine learning to improve the convergence speed and stability of the model.

[0200] Segmentation layer: In deep learning models, the role of the segmentation layer is to decompose the continuous video stream into individual static image frames for subsequent image-level analysis.

[0201] Convolutional layer: The component in a deep learning model responsible for extracting image features, using filters (also called convolution kernels) to identify local features and thus capture detailed information about the image.

[0202] Activation layer: Using nonlinear activation functions enables the model to learn and perform more complex tasks, which is an important means to improve deep learning performance.

[0203] Pooling layer: A layer used to reduce feature dimensions and extract main features. The maximum pooling method is usually used to extract the most significant features of the region.

[0204] Fully connected layer: A layer in a deep learning network in which each neuron is connected to all neurons in the previous layer to integrate all local features extracted previously.

[0205] Softmax function: A normalized exponential function, commonly used in the output layer of a multi-classification model, which can give probability scores belonging to each category.

[0206] Application scenarios:

[0207] This technical solution can be applied to situations where it is necessary to automatically identify and analyze abnormal behaviors in surveillance videos, such as urban security monitoring, traffic flow analysis, retail store theft detection, airport or station monitoring, etc.

[0208] Furthermore, taking the city public security monitoring system as an example, the system needs to automatically detect abnormal behaviors in the monitoring video, such as fighting, chasing and other possible public security incidents:

[0209] Video stream normalization processing:

[0210] The cloud server receives the decrypted video stream file and normalizes it through the input layer, making the subsequent analysis more stable and efficient.

[0211] Video stream segmentation:

[0212] The normalized video stream is processed through the segmentation layer and decomposed into a series of static images.

[0213] Feature extraction:

[0214] For each static picture, multiple filters on the convolution layer are used to extract features to obtain a first image feature set.

[0215] Non-Linear Learning:

[0216] The first image feature set is processed by applying a nonlinear function through an activation layer to enhance the expression ability of the model, thereby obtaining a second image feature set.

[0217] Feature Pooling:

[0218] The second image feature set is max-pooled using a pooling layer, thereby reducing the number of features and retaining the most important information, thereby obtaining a more abstract third image feature set.

[0219] Feature flattening and classification:

[0220] The fully connected layer integrates all features and flattens the pooled feature map into a vector form to prepare for classification decisions.

[0221] Output decision:

[0222] Finally, the softmax function is used to perform probability scoring on the feature vector, output the probability of abnormal behavior at each moment in the video stream, and generate an abnormal monitoring scenario analysis report.

[0223] After the implementation of this technical solution, the following beneficial effects can be derived:

[0224] Automated abnormal behavior detection: Using deep learning models, it is possible to automatically identify abnormal events in videos and reduce human resource requirements.

[0225] Improve detection accuracy: With the powerful capabilities of convolutional layers, pooling layers, and fully connected layers in image feature extraction and classification, the accuracy of abnormal behavior detection is improved.

[0226] Accelerate decision-making response time: By reducing the time required for automated analysis and report generation, a quick response can be made when abnormal events occur.

[0227] Improve the efficiency of the monitoring system: Due to the automated processing flow, a large amount of monitoring data can be processed effectively in real time, significantly improving the efficiency of the monitoring system.

[0228] Increase monitoring scene coverage: The model can be seamlessly expanded to multiple cameras and monitoring scenes to achieve wide-area coverage.

[0229] Another embodiment of the remote monitoring method of a camera in an embodiment of the present invention includes: configuring each intelligent monitoring camera in the composite monitoring network at least includes assigning an IP address to each intelligent camera, registering each intelligent camera to the management platform of the monitoring center, and setting monitoring parameters for each intelligent camera.

[0230] The above describes the remote monitoring method of the camera in the embodiment of the present invention. The following describes the remote monitoring device of the camera in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, a remote monitoring device for a camera includes:

[0231] The acquisition module is used to acquire video images of multiple monitoring scenes through a preset composite monitoring network, and encode each of the video images to obtain a corresponding video stream file; wherein the composite monitoring network is composed of a plurality of intelligent monitoring cameras connected;

[0232] A classification module, used to classify the video stream file to obtain a first data segment and a second data segment; wherein the first data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where an emergency occurs; and the second data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where no emergency occurs;

[0233] An encryption module is used to encrypt the first data segment and the second data segment respectively to obtain the encrypted first data segment and the encrypted second data segment, randomly generate a plurality of different obfuscated data segments, and combine the plurality of different obfuscated data segments, the encrypted first encrypted data, and the encrypted second encrypted data according to a preset arrangement rule to obtain an encrypted video stream file, and upload the encrypted video stream file to a cloud server;

[0234] A decryption module, used for the cloud server to decrypt the encrypted video stream file to obtain a decrypted video stream file; wherein the cloud server pre-stores a decryption algorithm that is the inverse of the encryption algorithm used to encrypt the first data segment and the second data segment;

[0235] The detection module is used to detect the decrypted video stream files based on the deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report.

[0236] The present invention also provides a remote monitoring device for a camera, the remote monitoring device for the camera comprising a memory and a processor, the memory storing computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor executes the steps of the remote monitoring method for the camera in the above-mentioned embodiments.

[0237] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the remote monitoring method of the camera.

[0238] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0239] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0240] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A remote monitoring method of a camera, characterized in that: The following steps are involved: The video images of multiple monitoring scenes are collected through a preset composite monitoring network, and each of the video images is encoded and processed respectively to obtain a corresponding video stream file; wherein the composite monitoring network is composed of a plurality of intelligent monitoring cameras connected; wherein the intelligent monitoring camera includes a remote image sensor, a video encoder, and an adaptive video stream dynamic adjustment module; the video images of multiple monitoring scenes are collected through a preset composite monitoring network, and each of the video images is encoded and processed respectively to obtain a corresponding video stream file, including: configuring each intelligent monitoring camera in the composite monitoring network to obtain a plurality of configured intelligent monitoring cameras; collecting video images of multiple monitoring scenes in real time through a remote image sensor, and converting the collected video images into digital signals by the remote image sensor to obtain a video data stream; encoding the video data stream according to a preset video encoding standard through a video encoder to obtain an encoded video data stream; dynamically adjusting the bit rate and picture quality parameters of the encoded video data stream according to real-time network changes through an adaptive video stream dynamic adjustment module using a machine learning algorithm and a prediction model to obtain a corresponding video stream file; The video stream file is classified to obtain a first data segment and a second data segment; wherein the first data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where an emergency occurs; and the second data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where no emergency occurs; Encrypting the first data segment and the second data segment respectively to obtain an encrypted first data segment and an encrypted second data segment, randomly generating a plurality of different obfuscated data segments, and combining the plurality of different obfuscated data segments, the encrypted first encrypted data, and the encrypted second encrypted data according to a preset arrangement rule to obtain an encrypted video stream file, and uploading the encrypted video stream file to a cloud server; The cloud server decrypts the encrypted video stream file to obtain a decrypted video stream file; wherein the cloud server pre-stores a decryption algorithm that is the inverse of the encryption algorithm used to encrypt the first data segment and the second data segment; The decrypted video stream file is detected based on the deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report, wherein the decrypted video stream file is detected based on the deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report, including: inputting the decrypted video stream file into the deep learning model; wherein the deep learning model at least includes an input layer, a segmentation layer, a convolutional layer, an activation layer, a pooling layer, a fully connected layer, and an output layer; the decrypted video stream file is normalized through the input layer to obtain a normalized video stream; the decrypted video stream file is normalized through the segmentation layer The normalized video stream is segmented to obtain multiple static images; the features of each static image are extracted by the multiple filters on the convolution layer to obtain a first image feature set; the image feature set is learned by the nonlinear activation function on the activation layer to obtain a second image feature set; the maximum pooling is performed on each second image feature region in the second image feature set by the pooling layer to obtain a third image feature set; the third image feature set is flattened by the fully connected layer to obtain each feature representation after classification; the softmax function on the output layer is used to output the decision of each feature representation after classification to obtain an abnormal monitoring scene analysis report; wherein the abnormal monitoring scene analysis report at least includes a probability score of the abnormal monitoring scene appearing in the video stream at each moment.

2. The remote monitoring method of a camera according to claim 1, characterized in that: The step of encrypting the first data segment and the second data segment respectively to obtain the encrypted first data segment and the encrypted second data segment comprises: Based on a preset data sensitivity assessment rule, a sensitivity assessment is performed on the first data segment to obtain a first encryption requirement for the first data segment. Based on a preset compliance assessment rule, a compliance assessment is performed on the second data segment to obtain a second encryption requirement of the second data segment; Based on the first encryption requirement of the first data segment, a first encryption algorithm matching the first encryption requirement is selected from a preset encryption algorithm database, and the first data segment is encrypted by the first encryption algorithm to obtain an encrypted first data segment; Based on the second encryption requirement of the second data segment, a second encryption algorithm matching the second encryption requirement is selected from a preset encryption algorithm database, and the second data segment is encrypted by the second encryption algorithm to obtain an encrypted second data segment.

3. The remote monitoring method of a camera according to claim 1, characterized in that: The configuration of each smart monitoring camera in the composite monitoring network at least includes allocating an IP address to each smart camera, registering each smart camera to the management platform of the monitoring center, and setting monitoring parameters for each smart camera.

4. A remote monitoring device for a camera, characterized in that: The remote monitoring device of the camera comprises: A collection module is used to collect video images of multiple monitoring scenes through a preset composite monitoring network, and encode each of the video images to obtain corresponding video stream files; wherein the composite monitoring network is composed of a plurality of intelligent monitoring cameras connected together; wherein the intelligent monitoring camera includes a remote image sensor, a video encoder, and an adaptive video stream dynamic adjustment module; collecting video images of multiple monitoring scenes through a preset composite monitoring network, and encoding each of the video images to obtain corresponding video stream files, including: configuring each intelligent monitoring camera in the composite monitoring network to obtain a plurality of configured intelligent monitoring cameras; collecting video images of multiple monitoring scenes in real time through a remote image sensor, and converting the collected video images into digital signals by the remote image sensor to obtain a video data stream; encoding the video data stream according to a preset video encoding standard through a video encoder to obtain an encoded video data stream; dynamically adjusting the bit rate and picture quality parameters of the encoded video data stream according to real-time network changes through an adaptive video stream dynamic adjustment module using a machine learning algorithm and a prediction model to obtain a corresponding video stream file; A classification module, used to classify the video stream file to obtain a first data segment and a second data segment; wherein the first data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where an emergency occurs; and the second data segment is the video stream data recorded by the intelligent monitoring camera in the monitoring scene where no emergency occurs; An encryption module is used to encrypt the first data segment and the second data segment respectively to obtain the encrypted first data segment and the encrypted second data segment, randomly generate a plurality of different obfuscated data segments, and combine the plurality of different obfuscated data segments, the encrypted first encrypted data, and the encrypted second encrypted data according to a preset arrangement rule to obtain an encrypted video stream file, and upload the encrypted video stream file to a cloud server; A decryption module, used for the cloud server to decrypt the encrypted video stream file to obtain a decrypted video stream file; wherein the cloud server pre-stores a decryption algorithm that is the inverse of the encryption algorithm used to encrypt the first data segment and the second data segment; The detection module is used to detect the decrypted video stream file based on the deep learning model deployed on the cloud server to obtain an abnormal monitoring scene analysis report, wherein the deep learning model deployed on the cloud server detects the decrypted video stream file to obtain an abnormal monitoring scene analysis report, including: inputting the decrypted video stream file into the deep learning model; wherein the deep learning model at least includes an input layer, a segmentation layer, a convolution layer, an activation layer, a pooling layer, a fully connected layer, and an output layer; normalizing the decrypted video stream file through the input layer to obtain a normalized video stream; The segmentation layer segments the normalized video stream to obtain multiple static images; the multiple filters on the convolution layer extract features of each static image to obtain a first image feature set; the image feature set is learned by the nonlinear activation function on the activation layer to obtain a second image feature set; the pooling layer performs maximum pooling on each second image feature region in the second image feature set to obtain a third image feature set; the fully connected layer flattens the features of the third image feature set to obtain each feature representation after classification; the softmax function on the output layer makes a decision output on each feature representation after classification to obtain an abnormal monitoring scene analysis report; wherein the abnormal monitoring scene analysis report at least includes a probability score of the abnormal monitoring scene appearing in the video stream at each moment.

5. A remote monitoring device for a camera, characterized in that: The remote monitoring device of the camera comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instruction in the memory so that the remote monitoring device of the camera executes the remote monitoring method of the camera as described in any one of claims 1-3.

6. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the remote monitoring method of the camera as described in any one of claims 1 to 3 is implemented.

Images