At frame-level, when anomaly score crosses a fixed threshold, then frames are treated as anomalous. But, how to localize the anomaly in spatial domain (with in anomalous frame)? Also how to make the threshold adaptive one?
Dear Rashmiranjan Nayak,
I suggest you to see links among these links you have yours and attached files on topic.
Article Unmasking the abnormal events in video
http://cucis.ece.northwestern.edu/projects/DMS/publications/AnomalyDetection.pdfhttps://www.researchgate.net/publication/317088245_Unmasking_the_abnormal_events_in_video
https://etheses.whiterose.ac.uk/22443/1/Anomaly%20detection%20in%20Video%20-%20Hanh%20Tran.pdf
https://www.sciencedirect.com/science/article/abs/pii/S0262885620302109?dgcid=rss_sd_all
https://openaccess.thecvf.com/content_CVPR_2020/papers/Pang_Self-Trained_Deep_Ordinal_Regression_for_End-to-End_Video_Anomaly_Detection_CVPR_2020_paper.pdf
Article An Efficient and Robust Unsupervised Anomaly Detection Metho...
Article Detection and Localization of Anomalous Motion in Video Sequ...
Best regards
Dear Rashmiranjan Nayak
These links might be an asset, have a look:
1. https://www.sciencedirect.com/science/article/abs/pii/S0262885620302109?dgcid=rss_sd_all
2. https://arxiv.org/abs/2008.08944
3. https://dl.acm.org/doi/10.1145/3394171.3413973
Kind Regards
Qamar Ul Islam
- Badges
- Science topic
- Similar topics
- Biological Science
- Histology
- Fixatives