An End to End Learning based Ego Vehicle Speed Estimation System

Estimating speed of a vehicle from only the video captured from within the vehicle without relying on any other supplementary information such as GPS coordinates or LIDAR data, is a challenging task. A deep neural network-based technique is proposed in this work that uses only the recordings from onboard dash camera, to estimate speed of the ego car. Three models are designed using various combinations of LSTM and CNN and evaluated on two benchmark datasets. From the results, it is observed that the proposed systems are well capable of generating estimates of the ego vehicle speeds with good accuracy