We have precomputed bounding boxes the detector run using a stride = 8x8, 
scaleratio = sqrt(sqrt(sqrt(2))), and the images are padded with 8px, using
model1 and model2. 

For reference we have also computed the bouding boxes using the INRIA
Pedestrain Detector (as implemented in Dalal&Triggs,CVPR'05) run at the 
same scaleratio and a finer one 1.05. 

Run main.m to run the benchmarks using precomputed detections. It outputs 
the figure benchmark.png/fig. Inside benchmark_det.m turn on the DISPLAY
flag to see the detections. You will need to download the INRIA Person Dataset
images to see the results. (http://pascal.inrialpes.fr/data/human/)


--
Subhransu Maji
http://www.cs.berkeley.edu/~smaji

Alexander C. Berg
http://www.cs.columbia.edu/~aberg
