This is the source code to train and evaluate VQ-VAE on haptic stimuli explored by human participants.
The dataset can be downloaded from this link http://www.allpsych.uni-giessen.de/DeepTouch/DeepTouch.zip . The password for extraction is "AnnaMatteo".
The content of this zip file:
- gt (the directory containing touch durations explored by participants)
- img (the directory containing haptic stimuli, the map reliefs)
- all_imgs.txt (each line corresponds to one image and ground-truth. lines are comma separated: <gt/img_index>,<participant_number>,<trial_number>,<stimuli_index>.)
- gts.txt (list of png files in the gt folder)
- imgs.txt (list of png in the img folder)
We have run this code with PyTorch 1.5.1 but it should work with other versions as well.
Navigate to the "src" folder and execute the following command:
python main.py <DATA_DIR> --test_inds 1 --batch-size 2 --epochs 100 -k 64 -kl 64 --hidden 64
<DATA_DIR>is the path to the extracted dataset containing the above described folders.test_indsis the flag for stating which participant should be used for testing. There are in total 12 participants. For instance,--test_inds 1 8 12means the training pipeline excludes the data of three participants (1, 8, 12), and uses them for validation. This is useful for n-fold cross-validation.kcorresponds to the number of vectors in the embedding space.klcorresponds to the lenght of vectors in the embedding space.hiddencorresponds to the number of kernels in de/convolution layers.
The output is stored in "results" directory.
To see a list of all supporting flags execute python main.py -h
To visualise networks prediction navigate to the "src" folder and execute the following command:
python main.py <DATA_DIR> --pred <MODEL_PATH> --test_inds 1 8 12
<DATA_DIR>as described above.test_indsas described above.<MODEL_PATH>path of the saved checkpoitns (those analaysed by use are in "models" directory).
The output is stored in "results" directory.