Data Augmentations Used on Training Data

facebookresearch / AVT

Code release for ICCV 2021 paper "Anticipative Video Transformer"

Apache License 2.0

152 stars 28 forks source link

Data Augmentations Used on Training Data #6

Closed okay-okay closed 2 years ago

okay-okay commented 2 years ago

Hi, I'm wondering if you would be able to point out where in the repository you have defined data augmentations used on training data (if any)? For instance, in the kinetics dataloader: https://github.com/facebookresearch/SlowFast/blob/master/slowfast/datasets/kinetics.py they use TRAIN_JITTER_SCALES, TRAIN_CROP_SIZE, AUG.NUM_SAMPLE, etc. Thanks so much!

rohitgirdhar commented 2 years ago

Hi, Yes, we define the augmentations in conf/data/defaults.yaml, which are sometimes overridden in the experiment configs. They are initialized in the train script here.

okay-okay commented 2 years ago

Hi,

Where can I find the parameters you've used for data augmentations? I can't seem to find any experiment configs, besides some txt files in expts/ folder.