DEVELOPMENT OF DRIVER’S MOVEMENT OPTIMIZED TRACKING (DMOT) ALGORITHM USING EYE GAZE EXTRACTION

Machines involvement in anticipating human activity is an intricate research issue. The analysis is aimed to deal with an important matter of just how exactly to expect a driver's activity whilst driving and enhance the anticipation period. The objective with this study would be to examine the present deep learning framework for assistive driving. This paper is different from the current solutions in just two manners. It suggests a simplified frame working with the driver's interior video data and also develops that a Driver's Movement Optimized Tracking (DMOT) algorithm. Vast majority of the existent condition of this method is dependent on interior and out features of these vehicles. Secondly, the suggested work has a tendency to boost the image layout recognition by introducing a combination of spatio-temporal data insert points (STIPs) for tracking movement together side eye cuboids by actions of prediction using deep learning. The suggested DMOT algorithm monitors the driver's movement with STIPs from the input. Additionally, a Fast Eye Gaze algorithm monitors eye motions. The features extracted in STIP and eye gaze have been fused and examined by a recurrent neural network to boost the prediction period, hence giving several additional moments to expect the driver's correct action. The operation of this DMOT algorithm is contrasted with all the prior implementations and discovered that DMOT offers 35 percent progress with respect to expecting driver's actions over two newly suggested deep learning methods.