This dataset contains 4 images per weather across 4 weathers, includes their corresponding semantic segmentation masks.
This dataset contains 2 videos per weather across 4 weathers, including their corresponding instance segmentation masks. *Note that our VIS model takes image inputs, not video inputs, but we included the compiled videos as it is easier to visualize the data this way*.
This dataset contains 5 images per weather across 4 weathers, includes their corresponding semantic segmentation masks and LIDAR data.
Email cv4ad2025@gmail.com for any full datasets!