https://github.com/jacobgil/keras-cam

Keras implementation of class activation mapping
https://github.com/jacobgil/keras-cam

deep-learning deep-learning-visualization keras visualization

Last synced: about 1 month ago
JSON representation

Keras implementation of class activation mapping

• Host: GitHub
• URL: https://github.com/jacobgil/keras-cam
• Owner: jacobgil
• Created: 2016-08-19T18:52:39.000Z (over 8 years ago)
• Default Branch: master
• Last Pushed: 2017-08-05T02:53:50.000Z (almost 8 years ago)
• Last Synced: 2025-04-02T11:04:29.366Z (about 1 month ago)
• Topics: deep-learning, deep-learning-visualization, keras, visualization
• Language: Python
• Size: 448 KB
• Stars: 335
• Watchers: 10
• Forks: 104
• Open Issues: 10
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
## Keras implementation of class activation mapping

Paper / project page: http://cnnlocalization.csail.mit.edu

Paper authors' code with Caffe / matcaffe interface: https://github.com/metalbubble/CAM

Blog post on this repository: http://jacobcv.blogspot.com/2016/08/class-activation-maps-in-keras.html 

Checkpoint with person/not person weights: https://drive.google.com/open?id=0B1l5JSkBbENBdk95ZW1DOUhqQUE

![enter image description here](https://raw.githubusercontent.com/jacobgil/keras-cam/master/examples/mona_lisa.jpg)

This project implements class activation maps with Keras.

Class activation maps are a simple technique to get the image regions relevant to a certain class.

This was fined tuned on VGG16 with images from here: 

http://pascal.inrialpes.fr/data/human

The model in model.py is a two category classifier, used to classify person / not a person.

	python cam.py --model_path cam_checkpoint.hdf5 --image_path=image.jpg

    usage: cam.py [-h] [--train TRAIN] [--image_path IMAGE_PATH]

              [--output_path OUTPUT_PATH] [--model_path MODEL_PATH]

              [--dataset_path DATASET_PATH]

	optional arguments:

	  -h, --help            show this help message and exit

	  --train TRAIN         Train the network or visualize a CAM

	  --image_path IMAGE_PATH

	                        Path of an image to run the network on

	  --output_path OUTPUT_PATH

	                        Path of an image to run the network on

	  --model_path MODEL_PATH

	                        Path of the trained model

	  --dataset_path DATASET_PATH

	                        Path to image dataset. Should have pos/neg folders,

	                        like in the inria person dataset.

	                        http://pascal.inrialpes.fr/data/human/