Crafting and learning for image matching
Place: CVC Large Lecture Room Abstract: I will overview key components of the finding correspondences in widely separated views with the local features. The key applications are 3d reconstruction, visual localization and image retrieval. The talk will starts with the basics and then go to the recent approaches, including, but not limited to research I was involved in: MODS, HardNet, AffNet. Also I will present a bag of unknown tricks for the image matching. Short Bio: Computer vision researcher at Czech Technical University in Prague, ex-CTO of Clear Research. Co-founder of Ukrainian Research Group “Szkocka”. Co-founder of Eastern European Computer Vision Conference (EECVC). Kaggle Master.