Delaware-Temple Occlusion Contour Dataset

Occluding Contour (OC) plays important roles in many computer vision tasks. We designed two benchmark datasets with both color images and OC images. The first dataset (category-16) simulates the ideal occluding contours by manuallly picking OCs from normal edges maps (Canny edges). In contrast, the occluding contours in the second dataset (DT-OC5) are automatically computed from the OC-Cam.


  1. Category-16 OC Dataset, category-16.zip
  1. DT-OC5, DT-OC5.zip

Related Publication:
Jin Sun, Christopher Thorpe, Nianhua Xie, Jingyi Yu, and Haibin Ling. "Category Classification Using Occluding Contours", International Symposium on Visual Computing (ISVC), 2010.