Method for returning to position of safety helmet based on face detection frame

Abstract

The invention discloses a method for returning to the position of a safety helmet and recognizing the safety helmet based on a face detection frame. The method comprises: establishing a multivariate machine learning model which is based on the center coordinate, width and height information of the face frame and is related to the same kind of position information of the safety helmet; and learningto obtain an influence coefficient of the face frame and the safety helmet according to the primary face frame information and a related learning formula, and substituting the influence coefficient and the secondary face frame information into a linear correlation equation to solve to obtain a candidate position of the safety helmet. The problem that automatic identification of the safety helmetis very difficult due to challenge problems such as shielding, small size, few data sets, small object identification, safety helmet carrier identification and the like of the safety helmet in an actual scene is solved. And meanwhile, compared with the human face, the pedestrian has large shielding and flexible changes, the detection is difficult when crowds are dense, and evidence cannot be obtained without the human face. According to the invention, face detection in a multi-face monitoring scene is realized, whether a crowd wears the safety helmet is judged, and video frames are stored as evidences to give an alarm at the same time.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a method for returning the position of a safety helmet based on a face detection frame. Background technique [0002] The wearing of hard hats is very important for many scenes, especially production construction sites. Moreover, being hit by an object and falling from an object on a construction site is the most common type of fatal accident, accounting for more than 68% of the total. However, in the actual scene, the geographical distribution of personnel is wide, the environment is complex, the operation volume is large, and the safety supervision is limited. It is difficult to realize real-time safety management throughout the process. Therefore, it is necessary to develop an automatic helmet identification method. [0003] In the helmet recognition method, the automatic recognition of the helmet is very difficult due to the challenges of occlusion, scale, small data...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is: due to the challenges of occlusion, scale, small data set, small object recognition, and helmet carrier identification in the actual scene of the helmet, it is very difficult to automatically identify the helmet; Compared with pedestrians, there are large occlusion and flexible changes, and it is difficult to detect in dense crowds, and there is no face to collect evidence

Images