diff --git a/_data/authors.yml b/_data/authors.yml index 78b2387..4dd26ed 100644 --- a/_data/authors.yml +++ b/_data/authors.yml @@ -68,7 +68,7 @@ Gronberg: Grover: firstname: ["Divya", "D.", "D.G.", "Divya Grover"] org: Chalmers University - scholar: gxlPOTcAAAAJ + scholar: gxlPOTcAAAAJ Hagerman: firstname: ["David", "D.", "D. H.", "David Hagerman"] org: Chalmers University of Technology @@ -200,3 +200,21 @@ Tram: Widahl: firstname: ["Jenny", "J.", "J.W.", "Jenny Widahl"] org: Zenseact +Carlsson: + firstname: ["Oscar", "O.", "O.C.", "Oscar Carlsson"] + org: Chalmers University of Technology +Gerken: + firstname: ["Jan E.", "J.E.", "J.E.G.", "Jan E. Gerken"] + org: Chalmers University of Technology +Linander: + firstname: ["Hampus", "H.", "H. L.", "Hampus Linander"] + org: Chalmers University of Technology +Spiess: + firstname: ["Heiner", "H.", "H. S.", "Heiner Spieß"] + org: Technical University Berlin +Ohlsson: + firstname: ["Fredrik", "F.", "F. O.", "Fredrik Ohlsson"] + org: Umeå University +Persson: + firstname: ["Daniel", "D.", "D. P.", "Daniel Persson"] + org: Chalmers University of Technology diff --git a/_publications/heal-swin/heal-swin.md b/_publications/heal-swin/heal-swin.md new file mode 100644 index 0000000..077d81d --- /dev/null +++ b/_publications/heal-swin/heal-swin.md @@ -0,0 +1,23 @@ +--- +layout: publication +permalink: /publications/heal-swin/ +title: "HEAL-SWIN: A Vision Transformer On The Sphere" +venue: CVPR24 +authors: + - Carlsson + - Gerken + - Linander + - Spiess + - Ohlsson + - Petersson + - Persson + +date: 2023-06-14 00:00:00 +00:00 +code: https://github.com/JanEGerken/HEAL-SWIN +arxiv: https://arxiv.org/abs/2307.07313 +n_equal_contrib: 2 +thumbnail-img: heal-swin.png +--- + +# Abstract +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, resulting in a one-dimensional representation of the spherical data with minimal computational overhead. We demonstrate the superior performance of our model for semantic segmentation and depth regression tasks on both synthetic and real automotive datasets. Our code is available at [https://github.com/JanEGerken/HEAL-SWIN](https://github.com/JanEGerken/HEAL-SWIN). \ No newline at end of file diff --git a/_publications/heal-swin/heal.swin.png b/_publications/heal-swin/heal.swin.png new file mode 100644 index 0000000..e4d4e12 Binary files /dev/null and b/_publications/heal-swin/heal.swin.png differ