https://github.com/jbeomlee93/AdvCAM

Anti-Adversarially Manipulated Attributions for Weakly and Semi-Supervised Semantic Segmentation (CVPR 2021)
https://github.com/jbeomlee93/AdvCAM

advcam cvpr2021 weakly-supervised-learning weakly-supervised-segmentation

Last synced: about 1 month ago
JSON representation

Anti-Adversarially Manipulated Attributions for Weakly and Semi-Supervised Semantic Segmentation (CVPR 2021)

• Host: GitHub
• URL: https://github.com/jbeomlee93/AdvCAM
• Owner: jbeomlee93
• License: mit
• Created: 2021-03-08T08:36:28.000Z (almost 4 years ago)
• Default Branch: main
• Last Pushed: 2022-07-13T01:48:46.000Z (over 2 years ago)
• Last Synced: 2024-08-03T01:11:52.655Z (5 months ago)
• Topics: advcam, cvpr2021, weakly-supervised-learning, weakly-supervised-segmentation
• Language: Python
• Homepage:
• Size: 2.62 MB
• Stars: 124
• Watchers: 2
• Forks: 16
• Open Issues: 13
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• Awesome-Semi-Supervised-Semantic-Segmentation - Code

README

        
# Anti-Adversarially Manipulated Attributions for Weakly and Semi-Supervised Semantic Segmentation

Input Image   |  Initial CAM   | Successive Maps with adversarial climbing

:-------------------------:|:-------------------------:|:-------------------------:

![a](demo/2008_004430.jpg)  |  ![b](demo/2008_004430_noreg_c_idx_0_iter_0.jpg) | ![c](demo/2008_004430_gif.gif)

The implementation of Anti-Adversarially Manipulated Attributions for Weakly and Semi-Supervised Semantic Segmentation, Jungbeom Lee, Eunji Kim, and Sungroh Yoon, CVPR 2021. [[paper](https://arxiv.org/abs/2103.08896)]

# Installation

- We kindly refer to the offical implementation of [IRN](https://github.com/jiwoon-ahn/irn).

- This repository is tested on Ubuntu 18.04, with Python 3.6, PyTorch 1.4, pydensecrf, scipy, chaniercv, imageio, and opencv-python.

## Usage

#### Step 1. Prepare Dataset

- Download PASCAL VOC 2012 benchmark: [Download](https://drive.google.com/file/d/1e-yprFZzOYDAehjyMVyC5en5mNq6Mjh4/view?usp=sharing).

#### Step 2. Prepare pre-trained classifier

- Pre-trained model used in this paper: [Download](https://drive.google.com/file/d/1G0UkgjA4bndGBw2YFCrBpv71M5bj86qf/view?usp=sharing).

- You can also train your own classifiers following [IRN](https://github.com/jiwoon-ahn/irn).

#### Step 3. Obtain the pseudo ground-truth masks for PASCAL VOC train_aug images and evaluate them

```

bash get_mask_quality.sh

```

#### Step 4. Train a semantic segmentation network

- To train DeepLab-v2, we refer to [deeplab-pytorch](https://github.com/kazuto1011/deeplab-pytorch). However, this repo contains only COCO pre-trained model. We provide [ImageNet pre-trained model](https://drive.google.com/file/d/14soMKDnIZ_crXQTlol9sNHVPozcQQpMn/view?usp=sharing) for a fair comparison with the other methods.

## Acknowledgment

This code is heavily borrowed from [IRN](https://github.com/jiwoon-ahn/irn), thanks [jiwoon-ahn](https://github.com/jiwoon-ahn)!