For more information, checkout the project site [website] and the paper [PDF].

• Python 3.6
• PyTorch 1.1.0
• Numpy
• Sklearn

• USCD Ped2 [dataset]
• CUHK Avenue [dataset]
• ShanghaiTech [dataset]

These datasets are from an official github of "Future Frame Prediction for Anomaly Detection - A New Baseline (CVPR 2018)".

Download the datasets into dataset folder, like ./dataset/ped2/

• 02/04/21: We uploaded the codes based on reconstruction method, and pretrained wieghts for Ped2 reconstruction, Avenue prediction and Avenue reconstruction.

• The training and testing codes are based on prediction method
• Now you can implemnet the codes based on both prediction and reconstruction methods.
• The codes are basically based on the prediction method, and you can easily implement this as

git clone https://github.com/cvlab-yonsei/projects
cd projects/MNAD/code
python Train.py # for training

• You can freely define parameters with your own settings like

python Train.py --gpus 1 --dataset_path 'your_dataset_directory' --dataset_type avenue --exp_dir 'your_log_directory'

• For the reconstruction task, you need to newly set the parameters, e.g,, the target task and the weights of the losses.

python Train.py --method recon --loss_compact 0.01 --loss_separate 0.01 # for training

• Test your own model
• Check your dataset_type (ped2, avenue or shanghai)

python Evaluate.py --dataset_type ped2 --model_dir your_model.pth --m_items_dir your_m_items.pt

• For the reconstruction task, you need to set the parameters as

python Evaluate.py --method recon --alpha 0.7 --th 0.015 --dataset_type ped2 --model_dir your_model.pth --m_items_dir your_m_items.pt

• Test the model with our pre-trained model and memory items

python Evaluate.py --dataset_type ped2 --model_dir pretrained_model.pth --m_items_dir m_items.pt

• Download our pre-trained model and memory items
  [Ped2 Prediction]
  [Ped2 Reconstruction]
  [Avenue Prediction]
  [Avenue Reconstruction]

• Note that, you need to set lambda and threshold to 0.7 and 0.015, respectively, for the reconstruction task. See more details in the paper.

@inproceedings{park2020learning,
  title={Learning Memory-guided Normality for Anomaly Detection},
  author={Park, Hyunjong and Noh, Jongyoun and Ham, Bumsub},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={14372--14381},
  year={2020}
}