ramprs/grad-cam

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1668 stars · 237 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Official Torch (Lua) implementation of Grad-CAM, the ICCV 2017 gradient-weighted class activation mapping technique for producing visual explanations of CNN decisions. It generates heatmap localizations highlighting image regions important for a predicted class, supporting image classification and visual question answering models.

Use cases

• visualize which image regions a CNN uses for a classification decision
• generate grad-CAM heatmaps for VGG-16, VGG-19, or AlexNet Caffe models
• explain visual question answering model answers with localization maps
• debug CNN failure modes via visual explanations
• produce class-discriminative localization maps without retraining the model
• research interpretability of convolutional neural networks

When to choose

• you need the reference implementation of the Grad-CAM paper for reproducibility
• you work with legacy Torch/Lua and Caffe CNN models
• you want to explain classification or VQA predictions with gradient-based heatmaps

When to avoid

• you use PyTorch or TensorFlow - modern ports like pytorch-grad-cam are actively maintained
• you need a license-compliant dependency - this repo has no license
• you want ongoing support or new model architectures - the repo is unmaintained

Facets

library · maturity abandoned

machine-learning deep-learning image-processing computer-vision data-visualization deep-learning computer-vision machine-learning image-processing lua grad-cam interpretability explainable-ai heatmap visual-explanation torch cnn iccv-2017 vqa gpu linux macos

Member repositories

For agents

markdown · JSON · MCP: product_card(name="ramprs/grad-cam")