Closed VivaainNg closed 4 years ago
Just found out that "split_file" corresponds to the path for multithumos.json file, whereas "split" corresponds to whether a particular dataset belongs to either "training" or "testing" categories (for both 200 & 213 samples, respectively).
Hello. When it comes to feature extraction using i3d, we have to run the code in charades_dataset_full.py first, and only then run the the extract_features.py module right?
But what does "split_file" and "split" means in the charades_dataset_full.py?
self.data = make_dataset(split_file, split, root, mode)
self.split_file = split_file
Thanks in advance.