Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{178790,
        author = {Anvay Sharma and Rahul Bajaj and Paras Gunjyal},
        title = {Anomalyze: Hybrid GNN and Transformer for Crowd Anomalies},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {12},
        pages = {5164-5171},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=178790},
        abstract = {This research proposes a hybrid model integrating GNN and Transformer architectures to detect crowd anomalies in real-time. The GNN component models the interactions between individuals in the crowd, while the Transformer captures long-range dependencies across video sequences. The hybrid approach enables robust identification of abnormal crowd behaviors, such as sudden dispersal, panic, or the presence of obstacles. Evaluation on benchmark datasets such as UCSD and UMN demonstrates improved accuracy, efficiency, and generalization compared to traditional methods like CNN-LSTM and handcrafted feature-based models. This work provides a foundation for real-time surveillance systems that ensure public safety in crowded environments},
        keywords = {Crowd anomaly detection, Graph Neural Networks, Transformers, Hybrid Model, Deep Learning},
        month = {May},
        }