Face acquisition edge terminal system

Abstract

The human face acquisition edge terminal system comprises a video monitoring acquisition module which is responsible for acquiring video signals through a camera or other monitoring equipment; the algorithm center is responsible for coordinating and scheduling all related functions related to a face recognition algorithm; the abnormal model is used for detecting and predicting an abnormal mode in the system and ensuring that potential safety risks or system faults are found in time; the early warning analysis module is started and performs event analysis and response when the system recognizes an abnormal condition; the Web configuration platform is used for providing a graphical interface for an administrator so as to conveniently configure and manage the system; and the management platform is used as a core management platform of the system and is responsible for performing centralized management on operation, maintenance, monitoring and optimization of the whole face recognition system. The system can be applied to various personnel access scenes so as to help to solve the access problem of unfamiliar personnel and provide necessary help to prevent more serious problems.

Description

Technical Field

[0001] The present invention relates to the field of face recognition systems, and in particular to a face acquisition edge terminal system. Background Art

[0002] The face recognition system is an emerging biometric technology with face recognition technology as its core, and is a high-tech technology that is being tackled in the international scientific and technological field today. It widely uses regional feature analysis algorithms, integrates computer image processing technology with biostatistics principles, uses computer image processing technology to extract portrait feature points from videos, and uses biostatistics principles to analyze and establish mathematical models. However, the existing system still lacks in accurate face detection, deduplication, image transmission, and timely warning of identified strangers.

[0003] Therefore, it is necessary to provide a face acquisition edge terminal system to achieve the above purpose. Summary of the invention

[0004] The purpose of the present invention is to provide a face collection edge terminal system.

[0005] The present invention achieves the above object through the following technical solutions:

[0006] A face collection edge terminal system, comprising:

[0007] include:

[0008] Video surveillance acquisition module, which is responsible for collecting video signals through cameras or other monitoring devices;

[0009] The algorithm center is responsible for coordinating and scheduling all functions related to the face recognition algorithm;

[0010] Anomaly models, which are used to detect and predict abnormal patterns in the system, ensuring that potential security risks or system failures are discovered in a timely manner;

[0011] Early warning analysis module: When the system identifies an abnormal situation (such as an unfamiliar face, an intruder, etc.), the early warning analysis module will start to analyze and respond to the event;

[0012] Web configuration platform, which provides administrators with a graphical interface to facilitate system configuration and management;

[0013] The management platform, as the core management platform of the system, is responsible for the centralized management of the operation, maintenance, monitoring and optimization of the entire face recognition system.

[0014] Furthermore, the video surveillance acquisition module is usually deployed in a field environment to obtain real-time images or video streams of the human body; extract clear facial images from the video stream; and continuously provide real-time or near real-time video data for subsequent processing modules to analyze.

[0015] Data transmission: The collected video or image data is transmitted to the backend system, or preliminary processing is performed on the edge device.

[0016] Furthermore, the algorithm center includes not only the core recognition algorithm, but may also include strategies such as optimization, calibration and tuning.

[0017] Furthermore, by analyzing abnormal situations that occur during the face recognition process (such as recognition failure, poor image quality, etc.), the abnormal model can help the system discover potential problems and correct them. It can also continuously monitor the system's operating status, detect any behavior that does not conform to the normal pattern, and trigger an alarm. In abnormal situations, the system can adjust the recognition strategy based on new data or environment to optimize the model's performance.

[0018] Furthermore, in real-time monitoring, if the early warning analysis module detects abnormal faces (such as unregistered persons) or abnormal behaviors (such as staying in a restricted area for a long time), the system will immediately trigger an early warning; rule engine: set predefined rules (such as face mismatch in a specific area, unauthorized persons entering a specific area, etc.), and perform early warning analysis based on these rules.

[0019] Furthermore, the Web configuration platform configures each module through the Web interface, such as adjusting the recognition threshold, setting alarm rules, modifying parameters, etc. Device management: manages the status and functions of connected cameras and terminal devices, and provides real-time data viewing and analysis functions to help administrators understand the system status.

[0020] Compared with the prior art, the present invention has broad application prospects for face collection, recognition and analysis based on video surveillance. The system can be applied to various scenes of people entering and exiting to help solve the problem of strangers entering and provide necessary assistance to prevent more serious problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0022] The user facial information, user device information, user personal information, etc., and data collected by the present invention, including but not limited to displayed data, analyzed data, etc., are information and data authorized by the user or fully authorized by all parties. The relevant data storage, use, processing, transmission, provision, disclosure, application, and processing comply with relevant laws, regulations, and standards, take necessary data confidentiality measures, do not violate process ethics, and have corresponding operation entrances for users to select, authorize or refuse.

[0023] See also Figure 1 This embodiment shows a face collection edge terminal system, including:

[0024] The video monitoring acquisition module 10 is responsible for collecting video signals through a camera or other monitoring equipment. It is usually deployed in a field environment to obtain images or video streams of the human body in real time.

[0025] Capture face images: Extract clear face images from video streams.

[0026] Provide video source: continuously provide real-time or near real-time video data for subsequent processing modules to analyze.

[0027] Data transmission: The collected video or image data is transmitted to the backend system, or preliminary processing is performed on the edge device.

[0028] The algorithm center 20 is responsible for coordinating and scheduling all functions related to the face recognition algorithm. It includes not only the core recognition algorithm, but also strategies such as optimization, calibration and tuning.

[0029] Face recognition algorithm: performs real-time face detection and matching, such as convolutional neural network (CNN) based on deep learning or other classic face recognition methods.

[0030] Load balancing: If the system involves multiple algorithm models or multiple processing units, the algorithm center will be responsible for the load balancing of the algorithms to ensure the efficient operation of the system.

[0031] Algorithm model update and optimization: Continuously improve and update the recognition algorithm to ensure the improvement of system performance and adapt to different environmental conditions.

[0032] Data preprocessing module 21: Before entering the algorithm processing, the original collected images or video streams often need to undergo some preprocessing to improve the accuracy and efficiency of recognition.

[0033] Image enhancement: Improve image quality by adjusting image brightness, contrast, sharpness, etc., making subsequent feature extraction and recognition more accurate.

[0034] Noise Removal: Remove background noise, especially in low light or complex environments, to reduce interfering signals and improve image recognizability.

[0035] Image cropping and normalization: Crop the face image to a suitable size, unify the image resolution, and normalize it to meet the model input requirements.

[0036] Face localization: locate and mark the face area in the image to provide accurate information for subsequent feature extraction.

[0037] The feature extraction module 22 extracts unique features that can represent the identity of the face from the pre-processed image for subsequent recognition or matching.

[0038] Facial feature extraction: Extract feature vectors from each face image using deep learning techniques (such as FaceNet, OpenFace, etc.). These feature vectors contain unique information about the face, such as the relative position, proportion, and shape of details such as the eyes, nose, and mouth.

[0039] Feature vector standardization: The extracted feature vectors are standardized to eliminate interference caused by environmental factors or shooting angles.

[0040] The model training module 23 is used to train the face recognition model through a large amount of training data so that it can accurately perform face recognition and matching.

[0041] Dataset construction and annotation: Collect annotated face image datasets and perform data cleaning, screening, and enhancement.

[0042] Train the model: Use deep learning or machine learning algorithms to train the model to learn the differences between different faces and the similarities between people.

[0043] Model optimization: Improve model performance by adjusting hyperparameters, changing network structure, and using different optimization algorithms.

[0044] Model evaluation: Evaluate the model through the test data set, calculate the accuracy, recall rate, F1 value and other indicators to ensure that the model has the ability to be applied in practice.

[0045] The anomaly model 30 is used to detect and predict abnormal patterns in the system to ensure timely detection of potential safety risks or system failures.

[0046] Abnormal pattern recognition: By analyzing abnormal situations that occur during the face recognition process (such as recognition failure, poor image quality, etc.), the model can help the system discover potential problems and correct them.

[0047] Real-time monitoring: Continuously monitor the system operation status, detect any behavior that does not conform to the normal pattern, and trigger an alarm.

[0048] Adaptive adjustment: In abnormal situations, the system can adjust the recognition strategy according to the new data or environment to optimize the performance of the model.

[0049] Early warning analysis module 40, when the system identifies an abnormal situation (such as an unfamiliar face, an intrusion into an area, etc.), the early warning analysis module will be activated and perform event analysis and response.

[0050] Anomaly detection: In real-time monitoring, if an abnormal face (such as an unregistered person) or abnormal behavior (such as staying in a restricted area for a long time) is detected, the system will immediately trigger an alert.

[0051] Rule Engine: Set predefined rules (such as face mismatch in a specific area, unauthorized personnel entering a specific area, etc.) and perform early warning analysis based on these rules.

[0052] The Web configuration platform 50 provides a graphical interface for administrators to configure and manage the system.

[0053] System configuration: Configure each module through the web interface, such as adjusting recognition thresholds, setting alarm rules, modifying parameters, etc.

[0054] Device management: manage the status and functions of connected cameras and terminal devices.

[0055] Data monitoring: Provides real-time data viewing and analysis functions to help administrators understand system status.

[0056] Warning push 51, when the system detects an abnormality, the system will send a warning message to the management personnel through message push.

[0057] Data analysis 52, analyzes and mines the data collected by the system to provide support for further decision-making and optimization.

[0058] Behavioral analysis: By analyzing historical face recognition data, we can identify the activity patterns or behavior patterns of people, such as the high frequency of people in a certain area and the flow of people.

[0059] Trend prediction: Based on the collected data, use statistical or machine learning methods to predict trends, such as predicting whether there is an abnormal flow of people in a certain area.

[0060] Visual display: Display data analysis results through charts, heat maps, etc., which is convenient for managers to view and understand.

[0061] The management platform 60, as the core management platform of the system, is responsible for the centralized management of the operation, maintenance, monitoring and optimization of the entire face recognition system.

[0062] User authority management: Set and manage user authorities, including operators, administrators, and ordinary users, to ensure that users with different authorities can access corresponding functions.

[0063] Logging and Auditing: Logs of all operations and events are recorded to facilitate system auditing and troubleshooting.

[0064] Alarm management and response: When an early warning or anomaly occurs, the management platform will summarize and present the alarm information to help administrators respond and handle it.

[0065] Report generation: Generate system reports and data analysis reports regularly and provide them to management for decision-making and optimization.

[0066] The present invention has broad application prospects for face collection, recognition and analysis based on video surveillance. The system can be applied to various scenes of people entering and exiting to help solve the problem of strangers entering and provide necessary assistance to prevent more serious problems. It has the following significant optimizations:

[0067] 1) Strong real-time performance: This system can analyze the real-time stream transmitted by the camera to promptly detect and warn of abnormal intrusion behavior.

[0068] 2) High accuracy: This system uses a machine learning model to strengthen and improve accuracy by continuously improving and supplementing personnel characteristics.

[0069] 3) Robustness: The system can work normally under a wide variety of conditions and is not easily disturbed or crashed.

[0070] The above are only some embodiments of the present invention. For those skilled in the art, several modifications and improvements can be made without departing from the creative concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A face collection edge terminal system, characterized in that: include: A video monitoring acquisition module (10), which is responsible for acquiring video signals through a camera or other monitoring equipment; Algorithm Center (20), responsible for coordinating and scheduling all functions related to face recognition algorithms; Anomaly models (30), used to detect and predict abnormal patterns in the system, ensuring timely detection of potential security risks or system failures; The early warning analysis module (40) is activated and performs event analysis and response when the system identifies an abnormal situation (such as an unfamiliar face, an intruder into an area, etc.); A Web configuration platform (50) provides a graphical interface for administrators to configure and manage the system; The management platform (60), as the core management platform of the system, is responsible for the centralized management of the operation, maintenance, monitoring and optimization of the entire face recognition system.

2. The face acquisition edge terminal system according to claim 1, characterized in that: The video monitoring acquisition module (10) is usually deployed in a field environment to obtain images or video streams of the human body in real time; extract clear facial images from the video stream; and continuously provide real-time or near real-time video data for subsequent processing modules to analyze. Data transmission: The collected video or image data is transmitted to the backend system, or preliminary processing is performed on the edge device.

3. The face acquisition edge terminal system according to claim 1, characterized in that: The algorithm center (20) includes not only the core recognition algorithm, but may also include strategies such as optimization, calibration and tuning.

4. The face collection edge terminal system according to claim 1, characterized in that: The anomaly model (30) can help the system identify potential problems and make corrections by analyzing anomalies that occur during the face recognition process (such as recognition failure, poor image quality, etc.). It can also continuously monitor the system's operating status, detect any behavior that does not conform to the normal pattern, and trigger an alarm. In abnormal situations, the system can adjust the recognition strategy based on new data or environment to optimize the model's performance.

5. The face collection edge terminal system according to claim 1, characterized in that: The early warning analysis module (40) will immediately trigger an early warning if an abnormal face (such as an unregistered person) or abnormal behavior (such as staying in a restricted area for a long time) is detected during real-time monitoring; the rule engine: sets predefined rules (such as a mismatch of faces in a specific area, an unauthorized person entering a specific area, etc.), and performs early warning analysis based on these rules.

6. The face collection edge terminal system according to claim 1, characterized in that: The Web configuration platform (50) configures each module through the Web interface, such as adjusting the recognition threshold, setting alarm rules, modifying parameters, etc. Device management: manages the status and functions of connected cameras and terminal devices, provides real-time data viewing and analysis functions, and helps administrators understand the system status.

Images