Dense Depth Estimation from Multiple 360-degree Images

In this project we want to generate disparity map and depth information from multiple 360-degree Images that can be helpful 3D scenes reconstruction.  I will focus on deep learning based depth map inference. I will model and train current state-of-the-art models with Tensorflow/Pytorch and compare their performance. I will convert the best performing model to ONNX representation which the DNN module (OpenVINO) can infer from. Thus integrating the model into libXcam for 360-degree depth map inference.