A method and system for completing the safety helmet shielding face area engine room inspection

CN122244798APending Publication Date: 2026-06-19HANSUN (SHANGHAI) MARINE TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HANSUN (SHANGHAI) MARINE TECH CO LTD
Filing Date
2026-03-25
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Traditional facial recognition technology has a low recognition rate in ship engine rooms due to factors such as helmet obstruction, complex lighting, and mechanical vibration, making it impossible to achieve efficient and reliable identity verification.

Method used

An Encoder-Decoder network is used for facial feature completion, and a GAN generator and Retinex algorithm are combined for illumination equalization. A training dataset adapted to the cabin scene is constructed for adversarial training to achieve end-to-end feature extraction and completion.

Benefits of technology

Achieving robust facial recognition in complex environments reduces human intervention and hardware requirements, improves recognition accuracy and management efficiency, and lowers operating costs and system complexity.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122244798A_ABST
    Figure CN122244798A_ABST
Patent Text Reader

Abstract

This application discloses a method and system for completing facial features in engine room inspections when a safety helmet obscures the face, relating to the field of ship inspection management. The method includes: acquiring facial images of inspection personnel inside the engine room; preprocessing the facial images to obtain preprocessed images; constructing and pre-training an Encoder-Decoder network containing a GAN generator; inputting the preprocessed image into the pre-trained Encoder-Decoder network to extract facial features from the preprocessed image and complete them into full facial features; performing similarity matching between the full facial features and a preset identity feature database to obtain a similarity matching level; and performing corresponding inspection management operations based on the similarity matching level. This application has the advantages of high recognition rate and strong robustness.
Need to check novelty before this filing date? Find Prior Art