Official code for the VideoGEM paper.
This repository consists of two main components:
- Description: Adaptation of the GEM model for video input with both dynamic and static weight application.
- Based on: GEM by Walid Bousselham
- Example Execution: Demonstrates the use of VideoClip as a video model on various datasets. Follow the instructions in testGEMViCLIP for execution.
- Additional Features:
- Prompt decomposition
- Verb and noun extraction as described in the VideoGEM paper