Computer Vision News - July 2021

Medical Computer Vision 25 Best of CVPR 2021 McGill University, gave an interesting talk entitled “ Seeing Through the Heart: Fiber Geometry at the Nanoscale ” on cardiac fiber geometry analysis and moving to micro scale to study cell membranes and sarcomeres, powered by classical computer vision techniques without using deep learning. Yuyin Zhou , Stanford University, gave an excellent talk highlighting clinical needs and key challenges of AI for medical imaging, and proposed to diagnose like a doctor by taking advantage of structural prior, data prior, and medical knowledge in real-world clinical applications. Mathias Unberath , Johns Hopkins University, gave a f antastic talk on quantitative endoscopy with a series of cutting-edge algorithms for extremely dense point correspondence via self- supervised learning, dense 3D reconstruction in endoscopy video, and ultimately towards dense monocular SLAM .