DRAD: A new model for dynamic real-time Avalanche detection from videos with residual depth-separable convolution and feature pyramid networks

Key Points

• Dynamic models improve avalanche detection accuracy using video analysis, ensuring timely alerts.
• A key metric shows increased detection efficiency, minimizing false alarms during real-world scenarios.
• Application of depth-separable convolution and feature pyramid networks strengthens the video-based approach.
• Implications suggest this technique could enhance safety in avalanche-prone areas, reducing risks.