arthurchen0518/DirectionNet

Where is the checkpoint and training set ?

Closed this issue · 2 comments

Hello! I'm trying to use your code for some application. In your instruction for training rotation, what are the <path_to_checkpoints_and_logs> and <path_to_training_set>? If I use "MatterportA_checkpoints", where are they?

Thanks a lot !!

<path_to_checkpoints_and_logs> is the directory you create to store the TensorFlow checkpoints and event files. <path_to_training_set> is the directory to the training data. "MatterportA_checkpoints" contains the TensorFlow checkpoints that we saved when we were training the model on the MatterportA dataset in the paper. Right now, due to the large size of the dataset, we haven't been able to release it. Our datasets are generated using generate_random_views in dataset.py from Matterport3D given the panorama images and camera labels. You can also use your own stereo dataset to train the model as long as the stereo images have the same intrinsic.

Oh ! I got it. Thanks for your help !!!