This folder contains the training, validation and testing splits of the AVCaps dataset.
The json files contain the audio, visual, audio-visual and the LLM-generated captions for each video file.

The videos along with these captions will soon be available freely on Zenodo after the anonymization phase is over.
 