Repository for (LGC->) CCL

July 18, 2025 · View on GitHub

Our new version has been accepted by ICCV2025. Code for paper "Salvaging the Overlooked: Leveraging Class-Aware Contrastive Learning for Multi-Class Anomaly Detection".

🛠️ Getting Started

Installation

  • Prepare general experimental environment 
    pip install -r requriements.txt

Dataset Preparation

Download datasets to data/ folder or set self.data.root in configs/lgc/lgc_data.py.

  • Real-IAD: A new large-scale challenging industrial AD dataset, containing 30 classes with totally 151,050 images; 2,000 ∼ 5,000 resolution; 0.01% ~ 6.75% defect proportions; 1:1 ~ 1:10 defect ratio.
  • BTAD: A real-world industrial anomaly dataset. The dataset contains a total of 2830 real-world images of 3 industrial products showcasing body and surface defects.
  • MVTec AD: It contains over 5000 high-resolution images divided into fifteen different object and texture categories.
  • VisA: It contains 12 subsets corresponding to 12 different objects as shown in the above figure. There are 10,821 images with 9,621 normal and 1,200 anomalous samples.
  • MANTA: It contains 38 categories and over 130K object-level images.

Train

  • Check data and model settings for the config file configs/model/model_data.py
  • Train with single GPU example: CUDA_VISIBLE_DEVICES=0 python run.py -c configs/lgc/lgc_data.py -m train
  • Train with multiple GPUs (DDP) in one node:
    • export nproc_per_node=8
    • export nnodes=1
    • export node_rank=0
    • export master_addr=YOUR_MACHINE_ADDRESS
    • export master_port=12315
    • python -m torch.distributed.launch --nproc_per_node=$nproc_per_node --nnodes=$nnodes --node_rank=$node_rank --master_addr=$master_addr --master_port=$master_port --use_env run.py -c configs/lgc/lgc_data.py -m train.
  • Modify trainer.resume_dir to resume training.

Test

  • Modify trainer.resume_dir or model.kwargs['checkpoint_path']
  • Test with single GPU example: CUDA_VISIBLE_DEVICES=0 python run.py -c configs/model/model_data.py -m test
  • Test with multiple GPUs (DDP) in one node: python -m torch.distributed.launch --nproc_per_node=$nproc_per_node --nnodes=$nnodes --node_rank=$node_rank --master_addr=$master_addr --master_port=$master_port --use_env run.py -c configs/lgc/lgc_data.py -m test.

Visualization

  • Modify trainer.resume_dir or model.kwargs['checkpoint_path']
  • Visualize with single GPU example: CUDA_VISIBLE_DEVICES=0 python run.py -c configs/lgc/lgc_data.py -m test vis=True vis_dir=VISUALIZATION_DIR

Checkpoints

Acknowledgement

Our benchmark is built on ADer and RD4AD, thanks their extraordinary works!

Citation

@article{fan2025salvaging,
          title={Salvaging the Overlooked: Leveraging Class-Aware Contrastive Learning for Multi-Class Anomaly Detection},
          author={Fan, Lei and Huang, Junjie and Di, Donglin and Su, Anyang and Song, Tianyou and Pagnucco, Maurice and Song, Yang},
          journal={arXiv preprint arXiv:2412.04769},
          year={2025}
        }