select images for codebook generation

gistfile1.txt

# -*- coding: utf8 -*-

import csv
import os


def listimgs(metafilepath, imgfilepath, n):
    #metafilepath: the path of the meta info of result
    #n: the number of a query to generate the codebook

    N = 355
    imgs = []
    for i in range(N):
        temp = []
        fn = os.path.join(metafilepath, 'meta_query_' + str(i) + '.dat')
        if os.path.exists(fn):
            with open(fn, 'rb') as csvf:
                info = csv.reader(csvf, delimiter=' ')
                for row in info:
                    temp.append([int(e) if j == 0 or j == 2 else e for (j, e) in enumerate(row)])
        sinfo = sorted(temp, key=lambda x: x[2])
        imgs.extend([os.path.join(imgfilepath, 'query_' + str(i), e[1].replace('textmeta.xml', 'imagethumb.jpg')) for e in sinfo[:n + 1]])
        with open('cblist.txt', 'w') as f:
            f.writelines('%s\n' % e for e in imgs)
        print len(imgs)
    return imgs

if __name__ == '__main__':
    listimgs('WebQueries\\metasByQuery', 'WebQueries\\imagesByQuery', 20)