pytorch-grad-cam
Python library that shows which parts of an image a computer vision model looked at to make its prediction.
- Stars
- 13k
- Forks
- 1.7k
- Licence
- MIT
- Language
- Python
- Last update
- 5 days ago
- Type
- Library
What it is
This is a Python package of explainability methods for computer vision models built with PyTorch. It produces heatmaps (CAMs) that highlight the image regions that most influenced a model's output, such as a classification label or an object detection result. People use it to check whether a model is looking at sensible features, either while developing a model or when diagnosing predictions in production.
- Generate a Grad-CAM heatmap over a ResNet50 classification prediction
- Compare GradCAM, AblationCAM and ScoreCAM outputs on the same image
- Visualize which regions drive a CLIP text prompt match
- Explain similarity between image embeddings
Great for
- Researchers comparing explainability methods for image models
- Engineers debugging CNN or Vision Transformer classifiers in PyTorch
- Teams working on object detection or semantic segmentation who need visual explanations
- Students learning how Grad-CAM-style methods work
Not ideal for
- People who need explanations for text or tabular models, since the README focuses on computer vision
- Users not comfortable writing Python code, because it is used through a programming API rather than a graphical interface
- Anyone needing a guarantee that an explanation is correct; the README includes metrics for checking trust but does not claim explanations are always faithful
How to get it
- 1
Install with a package manager
You'll need: Python · PyTorch
Install from PyPI
pip install grad-cam - 2
Run it
- Install the package with pip install grad-cam
- Load a PyTorch model and choose a target layer
- Import a method such as GradCAM and an image utility such as show_cam_on_image
- Run the CAM method on a preprocessed image and overlay the heatmap
- 3
Get the most out of it
- Try several CAM methods on the same input to compare results, as the README's examples do
- Use the metrics section of the README to check how trustworthy an explanation is
- Use the batched implementation for faster processing of multiple images
- Consult the linked documentation book for advanced tutorials
Downloads
Strengths
- Covers many methods (GradCAM, HiResCAM, ScoreCAM, EigenCAM and others) in one package
- Supports classification, object detection, segmentation and embedding similarity, per the README
- Released under the permissive MIT license
Trade-offs
- The README is truncated and does not show full API details or version requirements
- Some methods are expensive to run (the README notes a batched implementation for AblationCAM, implying the others are slower)
- No releases are published, so users depend on the latest repository state
How it compares
| Project | Stars | Type | Level | Updated | Best for |
|---|---|---|---|---|---|
| pytorch-grad-cam (this) | 13k | Library | intermediate | 5 days ago | Researchers comparing explainability methods for image models |
| annotated_deep_learning_paper_implementations | 67.6k | Learning resource | intermediate | 4 days ago | Students learning how transformers, GANs, or optimizers work internally |
| Made-With-ML | 49.9k | Learning resource | intermediate | 8 months ago | Software engineers who want to learn how ML models get deployed and maintained |
| pytorch-tutorial | 32.5k | Learning resource | intermediate | 3.2 years ago | Researchers or engineers who already know Python and want to learn PyTorch |
| shap | 25.8k | Library | intermediate | 4 days ago | Data scientists who need to explain tree-based models such as XGBoost, LightGBM, or CatBoost |
| pytorch_geometric | 24.1k | Library | intermediate | 15 days ago | Researchers who want ready-made implementations of published graph neural network methods |
Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.