Computer Vision News - November 2016

Qualitative results: Illustration of contour orientation learning. Row 1 shows the responses B k for 4 out of the 8 orientation bins. Row 2, from left to right: original image, contour strength, learned orientation map into 8 orientations, and hierarchical boundaries. Quantitative results: Evaluation metrics that were used are: standard F-measure at Optimal Dataset Scale (ODS) and Optimal Image Scale (OIS), as well as the Average Precision (AP), both evaluating boundaries (Fb) and regions (Fop) using the PASCAL VOC 2010 dataset. Computer Vision News Research 7 Research