Breakthrough in Measuring Snowfall Intensity with Hybrid Deep Learning and Surveillance Cameras

The sample snow streaks video dataset contains a total of 150 video segments, collected from three surveillance cameras with different backgrounds (50 clips from each camera). The snowflake extraction algorithm follows the method described in Section 2.1 of the manuscript. File-naming convention: For example, in the filename 002_6mm_0.3.avi: l 002 indicates the video ID; l 6mm represents the focal length of the camera used; l 0.3 denotes the snowfall intensity (mm/h).