Monkey Species Classifier — with Grad-CAM Interpretability

A VGG16 transfer-learning CNN classifies 10 monkey species from a photo. The right-hand panel shows the Grad-CAM heatmap — the regions of the image the model used to make its decision.

This is what interpretability looks like in practice: instead of trusting a black-box prediction, you can verify the model is looking at the monkey's features and not at random background details.

→ Full pipeline, training, tests on GitHub