PyTorch implementation of Luo et. al.: Efficient Deep Learning for Stereo Matching

Requires PyTorch 0.4

To train:

• Download the Kitti stereo dataset
• Edit KITTIPATH in config.py
• Edit AND create the save folder in config.py
• Run train.py A model is saved after every epoch

To test:

• Edit image paths in test.py
• Edit TEST_EPOCH in config.py to select the model to test
• Run test.py