Poselet work
================================
[File]seeds: 
	each line represents a seed patch, in the form of imgID#posInfo#cnt#kps#
	posInfo: [stx, sty, szx, szy] represents the sub-patch, the left corner of which is (stx, sty), and the size is (szx, szy)
	cnt: number of key points in the patch
	kps: the relative coordinates of key points in the patch
[Dir]seed_patches:
	all the seed patches extracted from images
[Dir]candidates:
	candidate patches in each image, the form of each file is the same as the form of 'seeds'
[Dir]collected_list:
	each file corresponds to the list of the closest candidate patches(the closest in each image) to each seed patch
	each line is in the form of imgID#posInfo#cnt#kps#scores
	scores:[dp, dv, d] corresponds the distance
[Dir]Sample:
	sample of the poselet training set
[Dir]img_no_dup:
	images with no duplications
[Dir]good:
	label jsons


================================
Python:
	See README in python