This project solved a 6D pose estimation problem in computer vision. The pipeline is to first use a segmentation network (e.g., U-Net) to segment all objects in an image, and then use a learning-based method, ICP (iterative closest point)to estimate the pose for each segmented object