diff.py create sequencial symlink, but some of pictures are skipped by diff value.

diff.py

#!/usr/bin/python

import cv2
import glob
import numpy as np
import os
import sys

"""
2019-02-18

calc diff compared with next jpg.
create symlinks with sequential names under ~/workspace/

the end of PATH directory should be HH, /00/ - /23/
every HH should have 3600 pictures.
pictures that are over 3600, will be skipped to create sequential symlinks.
e.g.
if 3602 pictures exist, diff.py make 2 groups.
diff.py find the smallest diff image and makes sequencial symlinks excluding the pic.

usage: python diff.py PATH
"""

# e.g. '/home/suzuki/2018-06-21/original/13/'
PATH = os.path.abspath(sys.argv[-1])
files = sorted(glob.glob(PATH + '*.jpg'))
count = 0
minimum = 1920 * 1080 * 255
print(PATH + 'mask.png')
mask = cv2.imread('mask.png', cv2.IMREAD_GRAYSCALE)
curr = cv2.imread(files[-1], cv2.IMREAD_GRAYSCALE) * mask

# TODO if-condition 3600
group_num = 3600 / (len(files) - 3600)

skipped = set()
print("path, diff value compared with prev img")
# find the smallest different image every group_num images
path = ""
for f in files:
  prev = curr
  curr = cv2.imread(f, cv2.IMREAD_GRAYSCALE) * mask
  diff = cv2.absdiff(prev, curr)
  diff_sum = diff.sum()
  print(f + ', ' + str(diff_sum))
  if minimum > diff_sum:
    minimum = diff_sum
    path = f
  count = count + 1

  if count > group_num:
    skipped.add(path)
    minimum = 1920 * 1080 * 255
    count = 0

print("skipped items: " + str(skipped))

# CHANGE this
count = int(PATH.split("/")[-2]) * 3600
for f in files:
    if f not in skipped:
        os.symlink(f, "~/workspace/" + str('{0:05d}'.format(count)) + ".jpg")
        count = count + 1