TRIBE Subcortex

TRIBE Subcortex extends the public TRIBE v2 demo into a focused Colab notebook for cortical prediction, deep-brain ROI activity, and subcortical response timelines.

Features

• Local video analysis in Google Colab
• Cortical TRIBE v2 prediction and surface visualization
• Subcortical prediction for response-related ROIs
• ROI dashboards for accumbens, amygdala, caudate, putamen, pallidum, thalamus, and hippocampus
• Composite subcortical response scores over time
• Hype vs release window comparison
• Peak timing tables
• Google Drive save/load for reusable prediction outputs

Notebook

• notebooks/tribe_subcortex.ipynb

Recommended Flow

1. Open notebooks/tribe_subcortex.ipynb in Google Colab.
2. Select a GPU runtime.
3. Run the install cell.
4. Restart the runtime after installation.
5. Run the HuggingFace login cell if gated models are needed.
6. Upload a video through the Colab file panel.
7. Recommended: rename the uploaded file to video.mp4, so the notebook path /content/video.mp4 works unchanged.
8. If you use another filename, update video_path = Path("/content/your_file.mp4") in the video cell.
9. Run cortical and subcortical prediction cells.
10. Save outputs to Google Drive before ending the runtime.

Uploading A Video

In Colab, open the left Files panel and upload your local video into /content. The simplest path is:

/content/video.mp4

Either rename your uploaded file to video.mp4, or edit the notebook cell:

video_path = Path("/content/your_file.mp4")

The /content folder is temporary. If the runtime restarts or disconnects, upload the video again or store reusable outputs in Google Drive.

Saved Runs

The notebook can save prediction outputs to Google Drive and load them later. This lets you refine plots, ROI windows, and dashboards without rerunning the full video encoding and prediction pipeline.

License

MIT