Where are the codes for processing sysu-30k dataset

[huge123]

I found the codes for processing the datasets like market1501, CUHK03, etc in data folder, but no sysu-30K. If you can tell how to process the proposed dataset, and how to set bagid, batchid, batchimage? Since there are no image-level labels, so how to sample batchimages?

[huge123]

Why the BagLabel in one bag contains duplicated numbers(id), and the image folders are also duplicated in bag_level_label.json? what are the relationships among the images in the same folder, they belong to the same id?

[wanggrun]

Let's take a part of bag_level_label.json as an example for illustration:
[
...
{"BagLabel": [3], "BagID": "bag0000000009", "BagImages": ["0000000003/*"]},
{"BagLabel": [2, 3, 4], "BagID": "bag0000000010", "BagImages": ["0000000002/*", "0000000003/*", "0000000004/*"]},
{"BagLabel": [1, 2, 3, 4, 5], "BagID": "bag0000000011", "BagImages": ["0000000001/*", "0000000002/*", "0000000003/*", "0000000004/*", "0000000005/*"]},
...
]

This indicates that all the images in ["0000000003/*"] can be assigned to a bag label [3]; and all the images in ["0000000002/*", "0000000003/*", "0000000004/*"] can be assigned to a bag label [2, 3, 4]; and all the images in ["0000000001/*", "0000000002/*", "0000000003/*", "0000000004/*", "0000000005/*"] can be assigned to a bag label [1, 2, 3, 4, 5].

[superaha]

I have a similar question:
Why there are duplicates in bags?

For example:

import json
data = json.load(open('bag_level_label.json'))
data[0]
{u'BagID': u'bag0000000000', u'BagLabel': [0], u'BagImages': [u'0000000000/']}
data[1]
{u'BagID': u'bag0000000001', u'BagLabel': [0, 0, 1], u'BagImages': [u'0000000000/', u'0000000000/', u'0000000001/']}
data[2]
{u'BagID': u'bag0000000002', u'BagLabel': [0, 0, 0, 1, 2], u'BagImages': [u'0000000000/', u'0000000000/', u'0000000000/', u'0000000001/', u'0000000002/*']}

why bag1 has two "0" and u'bag0000000002 even has three "0"?

Thank you.