CVPR Daily - Tuesday

3 DAILY CVPR Tuesday Bowen Cheng Instance segmentation is a computer vision task that locates objects in an image and delineates them with binary masks. When people train an instance segmentation model such as the Mask R-CNN model, it usually requires a data set with annotated masks of every object. You can spend around 80 seconds annotating a mask of an individual object, so the whole process takes a great deal of time. “ We try to reduce this annotation time by introducing a different form of annotation , ” Bowen tells us. “ We’re not trying to annotate the mask. Instead, for each object, we first annotate its bounding box. Doing this can take only seven seconds, which is more than five times faster than annotating the mask of every object. Then within this bounding box, we randomly sample a few points. We present each Bowen Cheng is a PhD student at the University of Illinois Urbana- Champaign under the supervision of Professor Alex Schwing. In August, he joins the Autopilot team at Tesla as a Research Scientist. He speaks to us ahead of his oral presentation today. Pointly-Supervised Instance Segmentation