Imageparsing.com (IP) is a web site affiliated with the Lotus Hill Institute (LHI)－an independent, non-profit, international research organization, established at EZhou, China, in 2005 by Dr. Song-Chun Zhu (Professor, Statistics and Computer Science, UCLA ) and a number of world renowned vision experts (Dr. Harry Shum, Professor Alan Yuille et al).
 
       In an effort to advance the field of computer vision, the sole objective of the IP website is to provide large scale annotated ground truth data to the general vision community. In two years, the Lotus Hill Institute has developed two proprietary modular image labeling tools -- ImageParser2.0 and VideoParser1.0 with easy to use interfaces and semi-automatic functions, and cultivated 30 well-trained and experienced full time labelers, recruited from local art colleges.
      
      The ground truth data set intends to cover almost all aspects of the computer vision and pattern recognition research: edges, contours, contour attributes, segmentation, grouping, occluded contour completion, text, object category recognition, scene, 3D world frames, UAV images, Google Earth images, video, and cartoon. The ground truth data are stored in a unified data structure – the And-Or graph representation and organized in a Database (MySQL) for retrieval and search. It has now over 3 million annotated object nodes by June, 2007.

      Part of the ground truth data are packed for various vision tasks and released one-by-one to the public in xml format together with matlab code to read and visualize the annotation.

      LHI also provides paid exclusive annotation service to research labs, companies, and researchers, and has done so successfully to a few clients. Our clients provide us their experimental data (images and video), specifications, and formats, we annotate the data accordingly and deliver the annotation results to the clients exclusively.