Search

CN-122024015-A - Portrait identification detection method and system based on AI

CN122024015ACN 122024015 ACN122024015 ACN 122024015ACN-122024015-A

Abstract

The invention discloses an AI-based portrait identification detection method and system, wherein the method comprises the steps of installing an AI portrait identification application according to a preset application environment configuration; acquiring shell files of current equipment, detecting whether the current equipment is provided with an AI dock file, pushing the AI dock file to the current equipment if the current equipment is not provided with the AI dock file, starting the AI dock file, entering the AI portrait identification application, carrying out portrait identification detection based on the AI dock file, and outputting a portrait identification detection result. The invention provides a human image recognition detection scheme which is light in weight, efficient in resources and robust in scene, not only improves the deployment feasibility and user experience in an operating system, but also improves the processing efficiency and detection accuracy of AI human image recognition.

Inventors

  • CHEN JIE

Assignees

  • 深圳开鸿数字产业发展有限公司

Dates

Publication Date
20260512
Application Date
20251217

Claims (10)

  1. 1. The human image recognition and detection method based on the AI is characterized by comprising the following steps of: Installing an AI image recognition application according to a preset application environment configuration; Acquiring shell files of current equipment, and detecting whether the current equipment is provided with an AI dock file or not; If the current equipment is not provided with the AI dock file, pushing the AI dock file to the current equipment; Starting the AI dock file, entering the AI image recognition application, and carrying out image recognition detection based on the AI dock file; And outputting a human image recognition detection result.
  2. 2. The AI-based portrait identification detection method according to claim 1, wherein the installing an AI portrait identification application according to a preset application environment configuration specifically includes: Acquiring authority configuration information of current equipment; judging whether the authority configuration information meets the installation requirement of the AI image recognition application or not; and if the installation requirement of the AI image recognition application is met, pushing and installing the HAP file corresponding to the AI image recognition application into the current equipment by using an hdc tool.
  3. 3. The AI-based portrait identification detection method as claimed in claim 2 wherein the rights configuration information includes camera access rights, storage read-write rights, network access rights, and AI reasoning rights.
  4. 4. The AI-based portrait identification detection method according to claim 1, wherein the obtaining a shell file of a current device and detecting whether the current device is equipped with an AI dock file specifically includes: acquiring shell files of the current equipment, and verifying the integrity of the shell files; If the shell file is complete, detecting the service availability of an AI dock; and if the AI docker service is not available, judging that the current equipment is not provided with the AI docker file.
  5. 5. The AI-based portrait identification detection method of claim 1, wherein the pushing the AI dock file to the current device specifically includes: Pushing the pre-constructed AI Docker mirror image to a Docker Hub warehouse; Installing a Docker runtime environment on the current device, starting the container and mapping the corresponding port.
  6. 6. The AI-based portrait identification detection method of claim 1, wherein the starting the AI dock file and entering the AI portrait identification application, and performing portrait identification detection based on the AI dock file, specifically includes: Executing a Docker pull command, and acquiring the AI Docker file from a Docker Hub warehouse; Verifying the integrity and version label of the AI dock file; Acquiring container configuration of the AI dock file, and starting a container corresponding to the AI dock file through a container starting command; checking the running state of the container and verifying whether the API service responds normally; and if the API service responds normally, calling a corresponding interface to run the AI image recognition application, and carrying out image recognition detection based on the AI dock file.
  7. 7. The AI-based portrait identification detection method of claim 6, wherein the invoking the corresponding interface runs the AI portrait identification application and performs portrait identification detection based on the AI dock file, and specifically includes: Uploading a face image or video stream to be detected through a RESTful API or a Web interface, and configuring a confidence threshold and an identification area; And loading a pre-training human image recognition model, and executing a human face detection, feature extraction and matching algorithm to obtain a human image recognition detection result, wherein the human image recognition detection result comprises human face coordinates, identity information and similarity information.
  8. 8. An AI-based portrait identification detection system, characterized in that the AI-based portrait identification detection system comprises: The application installation module is used for installing the AI image recognition application according to the preset application environment configuration; The file detection module is used for acquiring shell files of the current equipment and detecting whether the current equipment is provided with an AI dock file or not; The file pushing module is used for pushing the AI dock file to the current equipment when the current equipment is not provided with the AI dock file; The image recognition detection module is used for starting the AI dock file, entering the AI image recognition application, and carrying out image recognition detection based on the AI dock file; and the identification result output module is used for outputting a portrait identification detection result.
  9. 9. A terminal comprising a memory, a processor, and an AI-based portrait identification detection program stored on the memory and executable on the processor, which when executed by the processor, implements the AI-based portrait identification detection method of any of claims 1-7.
  10. 10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores an AI-based portrait identification detection program, which when executed by a processor, implements the steps of the AI-based portrait identification detection method according to any one of claims 1 to 7.

Description

Portrait identification detection method and system based on AI Technical Field The invention relates to the technical field of portrait identification, in particular to a portrait identification detection method and system based on AI. Background In the prior art, the implementation of portrait identification detection in an operating system faces multiple technical bottlenecks, and the high efficiency and the practicability of the portrait identification detection are obviously restricted. In particular, the current solutions have the following core problems in common: 1) The deployment complexity and the calculation efficiency are low, namely, the portrait identification system usually depends on a deep neural network architecture, such as a Convolutional Neural Network (CNN) -based model, and the model can realize high-precision feature extraction, but has huge algorithm calculation amount, so that the processing speed is obviously delayed. For example, in real-time video streaming processing, model reasoning can be time consuming beyond the system response threshold, failing to meet the timeliness requirements. In addition, the deployment process needs to integrate multiple components (such as camera driving, preprocessing modules and post-processing logic), so that the complexity of system integration is increased, and particularly, resource constraint is formed for embedded equipment. 2) The system resources are too high, and the high computing demands directly push up the CPU and GPU loads, so that the problems of memory and energy consumption are caused on mobile or edge computing equipment. Frequent model calls and data processing result in a slow system response, impaired user experience, manifested as application stuck or shortened battery life. Insufficient resource optimization further limits its popularity in low power scenarios. 3) The adaptability of the complex scene is insufficient, and the accuracy of the existing algorithm is sharply reduced under the condition of overlapping multiple images, low illumination or shielding. Illumination changes interfere with feature extraction, while facial occlusion (e.g., masks, glasses) results in loss of key points, affecting recognition robustness. Dynamic environments (e.g., mobile device shots) exacerbate recognition errors due to pose and expression changes, which can reduce the reliability of the system in real world applications. In view of the foregoing, there is a need for a lightweight, resource-efficient and scene-robust portrait identification detection scheme to improve the deployment feasibility and user experience in an operating system, and thus, the prior art needs to be improved. Disclosure of Invention The invention mainly aims to provide an AI-based portrait identification detection method and system, and aims to solve the problems of low processing efficiency and low detection accuracy in the prior art. In order to achieve the above object, the present invention provides an AI-based portrait identification and detection method, which includes the steps of: Installing an AI image recognition application according to a preset application environment configuration; Acquiring shell files of current equipment, and detecting whether the current equipment is provided with an AI dock file or not; If the current equipment is not provided with the AI dock file, pushing the AI dock file to the current equipment; Starting the AI dock file, entering the AI image recognition application, and carrying out image recognition detection based on the AI dock file; And outputting a human image recognition detection result. Optionally, the installing the AI-profile identification application according to the preset application environment configuration specifically includes: Acquiring authority configuration information of current equipment; judging whether the authority configuration information meets the installation requirement of the AI image recognition application or not; and if the installation requirement of the AI image recognition application is met, pushing and installing the HAP file corresponding to the AI image recognition application into the current equipment by using an hdc tool. Optionally, the permission configuration information comprises camera access permission, storage read-write permission, network access permission and AI reasoning permission. Optionally, the obtaining the shell file of the current device and detecting whether the current device is equipped with the AI dock file specifically includes: acquiring shell files of the current equipment, and verifying the integrity of the shell files; If the shell file is complete, detecting the service availability of an AI dock; and if the AI docker service is not available, judging that the current equipment is not provided with the AI docker file. Optionally, the pushing the AI dock file to the current device specifically includes: Pushing the pre-constructed AI Docker mirror im