Intelligent labeling method based on machine vision, electronic equipment, storage medium and program product

Abstract

The invention provides an intelligent labeling method based on machine vision. The intelligent labeling method comprises the following steps of camera detection, picture segment construction, optical flow calculation, RGB image matching, spatial relation binding, image feature combination, candidate word extraction, relevancy sorting and result labeling. Mainly based on an image retrieval and label generation technology, the method comprises the following steps: firstly, extracting candidate words by using text features, screening out a candidate word sequence in combination with a regression model, then establishing association between an image and a text, and finally, carrying out relevancy sorting on the candidate words according to a safety production standard model library so as to obtain an intelligent labeling result. The invention relates to an electronic device, a storage medium, and a program product. According to the intelligent labeling method based on machine vision, video stream data in a monitoring range is processed, key video frames are extracted, data labeling is carried out, follow-up processing of a follow-up system is facilitated, the recognition efficiency is high, the recognition accuracy is high, and the production safety requirements of enterprises can be met.

Description

technical field [0001] The invention relates to the technical field of production safety, in particular to an intelligent labeling method based on machine vision, electronic equipment, storage media and program products. Background technique [0002] In the field of safety production, unsafe behavior of people and unsafe stacking of objects are important factors that cause safety accidents. At present, for the monitoring of safety accident information, it still remains to manually identify whether there are unsafe behaviors of people and unsafe stacking of objects in the video captured by each monitoring unit. The efficiency and accuracy of identification are low, which cannot meet the needs of enterprises. production safety requirements. Therefore, there is an urgent need for an intelligent labeling method that recognizes "human unsafe behavior" and the unsafe stacking state of items through machine vision. Contents of the invention [0003] In order to overcome the def...

AI Technical Summary

Problems solved by technology

At present, for the monitoring of safety accident information, it is still stuck in manually identifying whether there are unsafe behaviors of people and unsafe stacking of objects in the video captured by each monitoring unit. The efficiency and accuracy of identification are low, which cannot meet the needs of enterprises. Production Security Requirements

Images