This directory contains example notebooks demonstrating the capabilities of the Panda model for LArTPC point cloud analysis.
Demonstrates feature extraction and linear probing using Panda's base encoder model.
Shows how to download and explore the PILArNet dataset using a HuggingFace integration.
Demonstrates semantic segmentation using Panda's fine-tuned model.
Shows panoptic segmentation combining semantic segmentation with instance segmentation. Provides two levels of clustering:
- Particle clustering: Groups points belonging to the same particle
- Interaction clustering: Groups points belonging to the same interaction
Each notebook is self-contained and can be run independently. The notebooks will automatically download required data from HuggingFace when needed.