Official Dataset for 📄 SoccerChat: Integrating Multimodal Data for Enhanced Soccer Game Understanding. This dataset supports multimodal research on soccer video understanding, especially in tasks involving natural language reasoning and event detection.

📁 Dataset Structure

The dataset is split into two partitions:

train (85,220 examples) validation (4,625 examples)

Each entry includes:

video: previewable video clip query: natural language… See the full description on the dataset page: https://huggingface.co/datasets/SimulaMet/SoccerChat.