We propose to integrate the reconstruction-based functionality into a novel self-supervised predictive architectural building block. The proposed self-supervised block is generic and can easily be incorporated into various state-of-the-art anomaly detection methods.

The open-access paper can be found at: https://arxiv.org/pdf/2111.09099.pdf

This code is released under the CC BY-NC-SA 4.0 license.

Our kernel is illustrated in the picture below. The visible area of the receptive field is denoted by the regions Ki, ∀i ∈ {1, 2, 3, 4}, while the masked area is denoted by M. A dilation factor d controls the local or global nature of the visible information with respect to M.

We provide implementation for both PyTorch and Tensorflow in the sspcab_torch.py and sspcab_tf.py scripts.

In order to work properly, you need to have a python version newer than 3.6 (we used the python 3.6.8 version).

If you use our block in your own work, please don't forget to cite us:

@inproceedings{Ristea-CVPR-2022,
  title={Self-Supervised Predictive Convolutional Attentive Block for Anomaly Detection},
  author={Ristea, Nicolae-Catalin and Madan, Neelu and Ionescu, Radu Tudor and Nasrollahi, Kamal and Khan, Fahad Shahbaz and Moeslund, Thomas B and Shah, Mubarak},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2022}
}

You can send your questions or suggestions to:

[email protected], [email protected]