The Research On Multiple Target Tracking Of Mining Heavy Trucks Based On Deep Learning

The rapid development of technology leads to the rapid progress of computer vision technology,and computer vision is receiving more and more attention.Computer vision is a discipline in which machines observe and think for themselves,and realize industrial automation and life facilitation by processing video images.With the development of deep learning research,how to apply deep learning to the traditional methods of computer vision has gradually become a topic of concern for scholars.In particular,the detection and recognition of heavy truck loads and heavy truck prediction and tracking in complex mining environment scenarios have some research value.Aiming at the problem of difficult detection and identification of heavy trucks under the influence of disturbing factors such as dust,fog and light transformation in the complex environment of mines.Based on the traditional detection of inter-frame differential method,optical flow method and background subtraction method,the fuzzy five-frame differential method is proposed to detect the moving targets,which effectively reduces the influence produced by numerous interference factors.In order to continuously track multiple targets in the video,an improved recurrent neural network model based on deep learning is used to predict and match multiple targets in the video in the spatio-temporal domain,which effectively reduces the influence of occlusion and can more accurately achieve the tracking of multiple moving targets.At the same time,considering the influence of various complex factors in the mining environment on the results,combined with the current state of development of technologies such as image processing and deep learning,deep learning methods are added to the traditional target detection,combined with the long and short-term memory model in deep learning to predict and track the wagons in the video in the spatio-temporal domain.It realizes the detection and tracking of target wagons on video timing and the effective suppression of various interferences in video images.This technology can better solve the problem of detecting and tracking multiple trucks in the complex environment of mines,and can be applied to many fields such as transportation and military.