Depth-and-Velocity-Prediction-using-Monocular-Images

Velocity and depth estimation from cameras has been an extensive area of research in Computer Vision. With the emergence of autonomous driving, it is understood to be a complicated yet imperative task which is still being explored both in industries and academia. The goal is to know the relative velocity and depth of a specific vehicle from a monocular camera mounted on our vehicle (referred to as ego-vehicle hereafter). In this project, we aim to leverage Neural Networks and Transformers to accomplish this task. This would also help cars respond better to unforeseeable triggers from the surroundings in comparison to humans. The code for this project is available at git code.