AnomLite: Efficient binary and multiclass video anomaly detection
Anomaly detection in video surveillance is critical for ensuring public safety, as manual monitoring of numerous video feeds is often challenging and prone to human error. Security operators can struggle to maintain focus over prolonged periods, leading to missed events or delayed responses. An auto...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-03-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025002506 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832582086592036864 |
|---|---|
| author | Anna K. Zvereva Mariam Kaprielova Andrey Grabovoy |
| author_facet | Anna K. Zvereva Mariam Kaprielova Andrey Grabovoy |
| author_sort | Anna K. Zvereva |
| collection | DOAJ |
| description | Anomaly detection in video surveillance is critical for ensuring public safety, as manual monitoring of numerous video feeds is often challenging and prone to human error. Security operators can struggle to maintain focus over prolonged periods, leading to missed events or delayed responses. An automated system can help mitigate this by providing continuous, reliable monitoring and rapid anomaly detection. In response to these challenges, we propose AnomLite, a lightweight yet effective model designed to detect anomalies in video streams through a hybrid architecture. The initial layers of MobileNetV2 are employed for efficient spatial feature extraction, capturing low-to-mid-level features such as edges and textures, while the last hidden state of an LSTM processes temporal dependencies to identify patterns indicative of anomalies. AnomLite is competitive due to its computational efficiency, requiring only 11 million parameters, and its robustness, achieving a ROC AUC of 0.99, Average Precision of 0.99 and F1-Score (Weighted) of 0.92 and outperforming comparable models in anomaly detection tasks. These attributes make AnomLite suitable for real-time applications in resource-constrained environments like urban centers and public transport hubs. The code could be found here: https://github.com/AnnaZverev/UCF_Crime.git |
| format | Article |
| id | doaj-art-4178c733910e4d14b4123507fed27500 |
| institution | Kabale University |
| issn | 2590-1230 |
| language | English |
| publishDate | 2025-03-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Results in Engineering |
| spelling | doaj-art-4178c733910e4d14b4123507fed275002025-01-30T05:14:52ZengElsevierResults in Engineering2590-12302025-03-0125104162AnomLite: Efficient binary and multiclass video anomaly detectionAnna K. Zvereva0Mariam Kaprielova1Andrey Grabovoy2Corresponding author.; Moscow Institute of Physics and Technology, Institutskiy Pereulok, 9, Dolgoprudny city, 141701, Moscow Oblast, Russian FederationMoscow Institute of Physics and Technology, Institutskiy Pereulok, 9, Dolgoprudny city, 141701, Moscow Oblast, Russian FederationMoscow Institute of Physics and Technology, Institutskiy Pereulok, 9, Dolgoprudny city, 141701, Moscow Oblast, Russian FederationAnomaly detection in video surveillance is critical for ensuring public safety, as manual monitoring of numerous video feeds is often challenging and prone to human error. Security operators can struggle to maintain focus over prolonged periods, leading to missed events or delayed responses. An automated system can help mitigate this by providing continuous, reliable monitoring and rapid anomaly detection. In response to these challenges, we propose AnomLite, a lightweight yet effective model designed to detect anomalies in video streams through a hybrid architecture. The initial layers of MobileNetV2 are employed for efficient spatial feature extraction, capturing low-to-mid-level features such as edges and textures, while the last hidden state of an LSTM processes temporal dependencies to identify patterns indicative of anomalies. AnomLite is competitive due to its computational efficiency, requiring only 11 million parameters, and its robustness, achieving a ROC AUC of 0.99, Average Precision of 0.99 and F1-Score (Weighted) of 0.92 and outperforming comparable models in anomaly detection tasks. These attributes make AnomLite suitable for real-time applications in resource-constrained environments like urban centers and public transport hubs. The code could be found here: https://github.com/AnnaZverev/UCF_Crime.githttp://www.sciencedirect.com/science/article/pii/S2590123025002506Anomaly detectionVideo surveillanceMobileNetV2LSTMDeep neural networksUCF-Crime |
| spellingShingle | Anna K. Zvereva Mariam Kaprielova Andrey Grabovoy AnomLite: Efficient binary and multiclass video anomaly detection Results in Engineering Anomaly detection Video surveillance MobileNetV2 LSTM Deep neural networks UCF-Crime |
| title | AnomLite: Efficient binary and multiclass video anomaly detection |
| title_full | AnomLite: Efficient binary and multiclass video anomaly detection |
| title_fullStr | AnomLite: Efficient binary and multiclass video anomaly detection |
| title_full_unstemmed | AnomLite: Efficient binary and multiclass video anomaly detection |
| title_short | AnomLite: Efficient binary and multiclass video anomaly detection |
| title_sort | anomlite efficient binary and multiclass video anomaly detection |
| topic | Anomaly detection Video surveillance MobileNetV2 LSTM Deep neural networks UCF-Crime |
| url | http://www.sciencedirect.com/science/article/pii/S2590123025002506 |
| work_keys_str_mv | AT annakzvereva anomliteefficientbinaryandmulticlassvideoanomalydetection AT mariamkaprielova anomliteefficientbinaryandmulticlassvideoanomalydetection AT andreygrabovoy anomliteefficientbinaryandmulticlassvideoanomalydetection |