Awesome

MoEAD

Code of MoEAD: A Parameter-efficient Model for Multi-class Anomaly Detection, accepted by ECCV 2024. article:https://www.ecva.net/papers/eccv_2024/papers_ECCV/html/11465_ECCV_2024_paper.php

Updates

• 26.09.2024: Code has been released.
• 20.11.2024: Some code mistakes have been corrected.

Abstract

Utilizing a unified model to detect multi-class anomalies is a promising solution to real-world anomaly detection. Despite their appeal, such models typically suffer from large model parameters and thus pose a challenge to their deployment on memory-constrained embedding devices. To address this challenge, this paper proposes a novel ViT-style multi-class detection approach named MoEAD, which can reduce the model size while simultaneously maintaining its detection performance. Our key insight is that the FFN layers within each stacked block (i.e., transformer blocks in ViT) mainly characterize the unique representations in these blocks, while the remaining components exhibit similar behaviors across different blocks. The finding motivates us to squeeze traditional stacked transformed blocks from N to a single block, and then incorporate Mixture of Experts (MoE) technology to adaptively select the FFN layer from an expert pool in every recursive round. This allows MoEAD to capture anomaly semantics step-by-step like ViT and choose the optimal representations for distinct class anomaly semantics, even though it shares parameters in all blocks with only one. Experiments show that, compared to the state-of-the-art (SOTA) anomaly detection methods, MoEAD achieves a desirable trade-off between performance and memory consumption. It not only employs the smallest model parameters, has the fastest inference speed, but also obtains competitive detection performance.

Overview of MoEAD

How to run

1. Installation

python version=3.8

pip3 install -r requirements.txt

Then we need to install fastmoe, a step-by-step tutorial for the installation procedure can be found here.

git clone https://github.com/laekov/fastmoe.git
cd fastmoe
python setup.py install

2. Dataset

• MVTec-AD

Download the MVTec-AD dataset from here. Unzip the file and move them to ./data/MVTec-AD. The MVTec-AD dataset directory should be as follows.

|-- data
  |-- MVTec-AD
​    |-- mvtec_anomaly_detection
​    |-- json_vis_decoder
​    |-- train.json
​    |-- test.json

• VisA

Download the VisA dataset from here. Follow the instruction in the link. After the part of data preparation in the instruction, move them to ./data/VisA_pytorch. The VisA dataset directory should be as follows.

|-- data
  |-- VisA_pytorch
​    |-- 1cls
​    |-- train.json
​    |-- test.json

3. Train

cd ./experiments/MVTec-AD/
sh train_torch.sh #NUM_GPUS #GPU_IDS

4. Test

cd ./experiments/MVTec-AD/
sh eval_torch.sh #NUM_GPUS #GPU_IDS

The evaluation results are shown the same as UniAD. In our paper, we use max for MVTec-AD and VisA.

Results and Checkpoints:

Acknowledgement

We thank for these great repositories including UniAD and fastmoe for assistance in our work.

Citation

If you find this paper and repository useful, please cite our paper.