2023 other ● stable

TrustNet

Explainable and trust-aware image classification prototype with uncertainty and OOD analysis.

Highlights

  • Implements uncertainty quantification and calibration for model confidence reliability.
  • Adds OOD detection via ODIN and Mahalanobis methods for trust-aware decisions.
  • Provides Grad-CAM visual explanations through an interactive Streamlit interface.

Related Projects