Video-based Seizure Detection — Docker Submission

This folder packages the trained 5-fold ST-GCN MIL model for the challenge Docker submission.

What the container does

• Reads test .npy pose segments from: /data/${INPUT}
• Writes prediction CSV to: /output/${OUTPUT}
• CSV columns: file_name,label

Expected runtime env vars

• INPUT: relative path (inside /data) to the folder containing test .npy files
• OUTPUT: relative path (inside /output) where the CSV should be written

Example: docker run --rm
-v /path/to/test:/data/test:ro
-v /path/to/out:/output
-e INPUT="test"
-e OUTPUT="pred.csv"