Deep Monocular Visual Odometry using PyTorch (Experimental)
Deep Monocular Visual Odometry implemented in PyTorch. This code is meant to be simple and easy to understand. The network architecture is inspired from DeepVO: Towards End-to-End Visual Odometry with Deep Recurrent Convolutional Neural Networks <https://arxiv.org/pdf/1709.08429>. Download and place the KITTI dataset in a desired location and run the code given the datapath argument to the dataset path (using --datapath).
You can modify the train/test sequences in dataset.py