grad-cam

Note:

1.The project works perfectly on Apple's Silicon M1. 2.Since the project did not process the image input, the input image must be 224x224 in size, otherwise the heat map effect is not ideal.

Project description

中文文档

• examples:save input images.
• results:save output images.

usage

1.build an environment

conda create -n your_env_name python==3.9

2.activate environment

• if your platform is PC:

activate your_env_name

• if your platform is Ubuntu or Mac:

source activate your_env_nam

3.install dependency

pip install -r requirements.txt

4.How to get grad_cam picture ？！

python cam.py --image-path path "for inputimage" --method "cam method"

for example

python get_cam.py --image-path './examples/both.png' --method 'gradcam'

example image

Reference

• https://github.com/jacobgil/pytorch-grad-cam
• https://zhuanlan.zhihu.com/p/371296750
• https://github.com/jacobgil/pytorch-grad-cam