Computer Vision Based Vehicle Intension Finding by Understanding Driver Hand Signal
Suraj Dhalwar, Sunidhi Ashtekar, Anuradha Pasupathy, Nagesh Poojary · 2021 IEEE Mysore Sub Section International Conference (MysuruCon) · 2021
Over a decade the increase in usage of vehicles and need for safety of people has greatly influenced the development of Advanced Driver Assistance Systems (ADAS) and Autonomous Driving (AD). The main purpose behind these systems is to enhance the ability of vehicle and its driver to react to various conditions on road. The ADAS provides high safety to vehicle drivers by warning about the external environment and possible threats. Similar to Vulnerable Road Users (VRUs) like pedestrian, cyclist, motorbike riders etc. if it is possible to anticipate the hand signals of the vehicle driver then the number of accidents can be controlled and its severity can be reduced drastically. This paper introduces a Computer Vision (CV) based solution to address this issue. A Convolutional Neural Network (CNN) architecture based automatic system is introduced that can help the ego-vehicle to recognize the hand signals of a vehicle driver and take the necessary actions in advance to prevent the road accidents. The proposed algorithm can recognize left, right and stop signals indicated by the vehicle driver that has never been addressed before. This method uses Regional Multi-Person Pose estimation (RMPE) - Alphapose [1] framework for human pose estimation and vehicle arm signal recognition algorithm is built on this framework. The testing of proposed work is done on self-captured dataset. The proposed work provides accurate results to recognize hand signals of the vehicle driver and the accuracy is 92.87% at 10 fps.