This page provides source codes, datasets, instructions for the ECCV'08 paper:
Zhe Lin and Larry S. Davis, "A pose-invariant descriptor for human detection and segmentation,", European Conference on Computer Vision, Marseille, France, 2008..
The provided datasets, codes may only be used for research purposes ONLY.
Source codes/Readme (c++ win32 console app):
Train/Test Codes: PidHumanDet.zip
Update(06/05/2009): PidHumanDet.zip is updated. Now, test programs can output a txt file of the detection result (i.e. # of boxes, box coordinates, confidence values).
Example:
Command line: $PidBoostTest ..\data\test\svmtestconf.txt (using SVM) or $PidBoostTest ..\data\test\cascadetestconf.txt (using cascade)
Example input and result images:
Note: please direct any inquires and questions to zhelin AT umiacs.umd.edu.
