arXiv:2506.03162v3 Announce Type: replace-cross Abstract: The rapid proliferation of surveillance cameras has increased the demand for automated violence detection. While CNNs and Transformers have shown success in extracting spatio-temporal features, they struggle with long-term dependencies and computational efficiency. We propose FuseMamba-VD: Dual Branch VideoMamba with Gated Class Token Fusion (GCTF), an efficient architecture combining a dual-branch design and a state-space model (SSM) backbone where one branch captures spatial features, while the other focuses on temporal dynamics. The
The proliferation of surveillance cameras and advancements in AI models like Mamba are creating an immediate need and opportunity for more efficient and accurate automated violence detection systems.
Improved automated violence detection can significantly enhance public safety and security by enabling real-time monitoring and rapid response to incidents, particularly in large-scale surveillance environments.
This advancement provides a more computationally efficient and effective method for identifying violent activities in video feeds, moving beyond the limitations of prior CNN and Transformer architectures.
- · Surveillance technology providers
- · Public safety agencies
- · Smart city developers
- · AI model developers
- · Criminals and violent actors
More widespread and effective deployment of automated violence detection in public and private spaces.
Potential societal debates around privacy implications due to enhanced surveillance capabilities.
The technology could be integrated into autonomous security systems, leading to predictive policing capabilities.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI