A self-learning approach is proposed towards solving scene-specific pedestrian detection problem without any human¡¯ annotation involved. The selflearning approach is deployed as progressive steps of object discovery, object enforcement, and label propagation.
This is a matlab code of Self-learning Scene-specific Pedestrian Detectors using a Progressive Latent Model. Copyright Reserved by University of Chinese Academy of Sciences. It is free for academy purpose. Please contacet [email protected] if you have more problems
Runtime enviroment: Matalb12 or later vergion,
-
Download the Edgebox proposal generation code from http://vision.ucsd.edu/~pdollar/research.html
-
Download the DPM code from Ross Grishick's UC berkely websit
-
Supose the video name is 'PETS09-S2L2.avi', put the video in the dataset 'data'
-
Make a folder as the name of video
-
Randomly prepare >1000 negtive images in the data\videoname\neg folder Prepare the neg_filelist.txt in the data\videoname foler
-
Run Demo by inputting st_learning('.\data\PETS09-S2L2.avi')