HEAL-SWIN: A Vision Transformer On The Sphere
This addresses a domain-specific problem for robotics applications such as autonomous driving by enabling efficient, distortion-free processing of spherical image data.
The authors tackled the problem of processing high-resolution wide-angle fisheye images for robotics by introducing HEAL-SWIN, a vision transformer on the sphere that avoids projection and distortion losses, demonstrating superior performance on synthetic and real automotive datasets for tasks like semantic segmentation, depth regression, and classification.
High-resolution wide-angle fisheye images are becoming more and more important for robotics applications such as autonomous driving. However, using ordinary convolutional neural networks or vision transformers on this data is problematic due to projection and distortion losses introduced when projecting to a rectangular grid on the plane. We introduce the HEAL-SWIN transformer, which combines the highly uniform Hierarchical Equal Area iso-Latitude Pixelation (HEALPix) grid used in astrophysics and cosmology with the Hierarchical Shifted-Window (SWIN) transformer to yield an efficient and flexible model capable of training on high-resolution, distortion-free spherical data. In HEAL-SWIN, the nested structure of the HEALPix grid is used to perform the patching and windowing operations of the SWIN transformer, enabling the network to process spherical representations with minimal computational overhead. We demonstrate the superior performance of our model on both synthetic and real automotive datasets, as well as a selection of other image datasets, for semantic segmentation, depth regression and classification tasks. Our code is publicly available at https://github.com/JanEGerken/HEAL-SWIN.