Each branch consists of one task of the homework
The contribution of the work
Task 1: Luu Tung Hai and Abdul Rehman Shah implemented the flow work of the YOLOv4, we have to label images again because of the limitation in coding, also in the annotation .txt file, it consists of 3D coordinates and using API's yolov4 requires only classes( car:1, pedestrian:2,..) and 2D coordinates.
Task 2: We tried to find the points belonging to a particular object but failed. we did manage to make the function to find the x,y from z coordinates from the depth images. we also defined a function to draw 3d box.
Task 3: Luu Tung Hai and Abdul Rehman Shah implemented some networks, metrics, then we compared and chose the best results.
Task 4: In this task
Luu Tung Hai and Abdul Rehman Shah implemented Ransac algorithm.
Extra task: Ahmed implemented some transformations using Opencv functions( used filters like Sobel operator and color space thresholding and histogram and sliding windows for line detection and prediction and lane curvature calculations)
Documentations: Abdul Rehman Shah , Luu Tung Hai and Pissolati