CMAL-Net

Code release for 'Learn from Each Other to Classify Better: Cross-layer Mutual Attention Learning for Fine-grained Visual Classification'. You may check more details in our paper published in Pattern Recognition if you are interested in our work.

Environment

This source code was tested in the following environment:

Python = 3.7.11

PyTorch = 1.8.0

torchvision = 0.9.0

Ubuntu 18.04.5 LTS

NVIDIA GeForce RTX 3080 Ti

Dataset

• (1) Download the FGVC-Aircraft, Stanford Cars, and Food-11 datasets and organize the structure as follows:

dataset folder
├── train
│   ├── class_001
|   |      ├── 1.jpg
|   |      ├── 2.jpg
|   |      └── ...
│   ├── class_002
|   |      ├── 1.jpg
|   |      ├── 2.jpg
|   |      └── ...
│   └── ...
└── test
    ├── class_001
    |      ├── 1.jpg
    |      ├── 2.jpg
    |      └── ...
    ├── class_002
    |      ├── 1.jpg
    |      ├── 2.jpg
    |      └── ...
    └── ...

• (2) modify the path to the dataset folders.

Dependencies

• (1) Inplace-ABN

Install Inplace-ABN following the instructions:

https://github.com/Alibaba-MIIL/TResNet/blob/master/requirements.txt

https://github.com/Alibaba-MIIL/TResNet/blob/master/INPLACE_ABN_TIPS.md

• (2) TResNet_L

Download the folder src from https://github.com/Alibaba-MIIL/TResNet and save it as:

Code
├── basic_conv.py
├── utils.py
├── train_Stanford_Cars_TResNet_L.py
├── ...
├── src

Training

Run the scripts for training, such as python train_Stanford_Cars_TResNet_L.py.

Acknowledgement

Part of the training code is inspired by Du et al, and many thanks to them.

Citation

@article{LIU2023109550,
title = {Learn from each other to Classify better: Cross-layer mutual attention learning for fine-grained visual classification},
journal = {Pattern Recognition},
volume = {140},
pages = {109550},
year = {2023},
issn = {0031-3203},
doi = {https://doi.org/10.1016/j.patcog.2023.109550},
url = {https://www.sciencedirect.com/science/article/pii/S0031320323002509},
author = {Dichao Liu and Longjiao Zhao and Yu Wang and Jien Kato}
}