Computer Vision for Science

Motivation

Our workshop aims to bring together people working at the intersection of computer vision and the sciences. While many computer vision researchers are working at the intersection of computer vision and a science discipline, these efforts are often not highlighted at CVPR. As a result researchers remain disconnected, unaware of each others’ work, and miss opportunities to learn from each other. Early career researchers walking the poster floor of CVPR can get the mistaken impression that computer vision researchers do not work on or care about these problems. 

On the other hand, there has been  a recent surge in interest in the community in these topics, with a CVPR 2023 keynote and ICCV 2023 keynote covering AI For Science. These have been large-scale affairs that preclude focused discussions between practitioners. Given the need and considerable interest, we believe that now is the time to start having a workshop on computer vision for science.

We aim to highlight work in this space and are interested in any topic that covers both computer vision and the sciences:

• Computer vision topics in this area often include (but are not limited to): reconstruction, recognition, segmentation and counting, human-in-the loop efforts, low-shot learning, domain adaptation and sim2real, video analysis, joint design of hardware and software. 

• Science topics include (but are not limited to): astrophysics via a variety of instrument types (radio, light, spectropolarimetry), chemistry, biology, neuroscience, and ecology.

Format

CV4Science will be a full day workshop incorporating:

• Talks from a set of senior researchers at the interface betwen computer vision and science, including both computer vision researchers and domain experts

• Posters from junior researchers that are selected on the basis of an extended abstract. These posters will be given a short spotlight plus a poster. 

We will not have long-form workshop papers.