HybridMamba combines visual transformers with state-space modeling to precisely locate crash events in lengthy surveillance videos, addressing a critical challenge in traffic safety monitoring.

• Efficiently processes extended video footage while maintaining temporal precision
• Uses multi-level token compression and hierarchical processing to manage computational costs
• Enables faster emergency response through accurate crash time identification
• Enhances infrastructure planning with detailed crash analytics

This innovation significantly improves security systems by enabling precise incident detection in surveillance footage, potentially reducing response times and saving lives in traffic emergencies.

Crash Time Matters: HybridMamba for Fine-Grained Temporal Localization in Traffic Surveillance Footage