CV Lab 04: Grad-CAM Interpretability

Use Grad-CAM to interpret model predictions on images.

What you'll do

  • Generate Grad-CAM heatmaps for a trained CNN
  • Interpret which image regions drove a prediction
  • Use interpretability to sanity-check a defect-detection model
Requirements: see requirements.txt in the Resources folder. Install with pip install -r requirements.txt after creating your virtual environment.
No bundled CSV download for this lab. Follow the notebook setup instructions.

Opening in Colab needs a free Google account — sign in to Google first.

← CV Lab 03: Model Comparison and Metrics Fruit and Vegetable Classification →