About Region Confusion Mechanism

Hi！it's great work! But I have some questions about the region confusion mechanism in the code (below). What's the mean of the parameter **pro** and **RAN**，and why **pro** should >=5, I can't find this information in your paper.

```
pro = 5
if pro >= 5:
tmpx = []
tmpy = []
count_x = 0
count_y = 0
k = 2
RAN = 2
for i in range(crop[1] * crop[0]):
tmpx.append(images[i])
count_x += 1
if len(tmpx) >= k:
tmp = tmpx[count_x - RAN:count_x]
random.shuffle(tmp)
tmpx[count_x - RAN:count_x] = tmp
if count_x == crop[0]:
tmpy.append(tmpx)
count_x = 0
count_y += 1
tmpx = []
if len(tmpy) >= k:
tmp2 = tmpy[count_y - RAN:count_y]
random.shuffle(tmp2)
tmpy[count_y - RAN:count_y] = tmp2
random_im = []
for line in tmpy:
random_im.extend(line)
```

Locate the Python code containing the shown region-confusion mechanism and read the paper discussed in the issue. Determine what pro and RAN represent and why the pro >= 5 condition is used; the issue is complete when those meanings and the threshold rationale are documented or explained.

Written by the indexing model from the issue text.