Computer Vision News - January 2020

2 Summary Challenge 18 year, around 50 people took part in a full-day program with two keynotes and plenty of interesting discussion and debate. There were three sub-challenges: Robust Medical Instrument Segmentation; Stereo Correspondence and Reconstruction of Endoscopic Data; and Surgical Workflow and Skill Analysis. Stefanie tells us that what’s missing from the data so far is multiple centers: “In the past, we’ve had sub-challenges that were focused on data from only one hospital, for example. It might be that their methods won’t generalize to a data set from another center. This is a challenge that we have to tackle. Stefanie Speidel is Professor for Translational Surgical Oncology at the National Center for Tumor Diseases in Dresden. Earlier this year she co-organized the Endoscopic Vision (EndoVis) challenge at MICCAI 2019 in Shenzhen. She speaks to us about this and her other work. EndoVis started out as an endoscopic vision CAI challenge at MICCAI 2015 . It was an open call to encourage groups to share their data because there was a feeling that everyone was struggling to get good annotated data for computer-assisted surgery. Stefanie with co-organizers Lena Maier-Hein and Danail Stoyanov hadn’t set their hopes high for a strong response but were surprised to receive a number of applications from groups who wanted to take part. Fast forward to 2019, and a poll at the Surgical Data Science Workshop voted EndoVis as one of the most useful providers of public data sets for surgical data science. At the MICCAI satellite event this with Stefanie Speidel