Projects

JTAG

JTAG is an application I wrote in the summer of 2003 that had the ability to manipulate, display, and classify rectangular sections of elements on each page.

Scripts

Input

Journal Region Ground Truth Dataset

I've hand labelled region information for 1204 pages of journal and conference articles. The articles are taken from the proceedings of NIPS, and PAMI, as well as several articles taken from JMLR

Region information is stored in flat text files and includes region boundary co-ordinates, a fine-grained region label, other JTAG specific information like learning algorithm used, whether the selection was cropped or not, how long it took to create the region, and whether any attempts were made to resize the region.

If you'd like a copy of this dataset (or associated matlab scripts for parsing the data), just email me.