CA_Market1501

Here is the official repository for CA_Market1501 attribute dataset. You can find more details on CA-Market: A Partially Categorical AnnotatingApproach Based on Market1501 Dataset for Attribute Detection

About dataset

We have annotate 45 attributes for Market-1501. The original dataset contains 751 identities for training and 750 identities for testing. The attributes are annotated in the image-based level on gt_bbox folder of Market-1501 dataset, thus the file contains a CA_Market_with_id.npy file with size of 25259 x 46 which includes 45 binary attributes of train and test images of sorted gt_bbox folder from Market-1501 dataset and their id. The size of train_idx is 12567 and the size of test_idx is 12692.

The 45 attributes are:

attribute	part	column index
gender	gender	0
cap	head	1
hairless	head	2
short hair	head	3
long hair	head	4
knot hair	head	5
head (colorful1, black0)	head color	6
Tshirt/shirt	body	7
coat	body	8
top	body 9
patterned	body type	10
white	body color	11
red	body color	12
yellow	body color	13
green	body color	14
blue	body color	15
gray	body color	16
purple	body color	17
black	body color	18
backpack	bags	19
bag	bags	20
no bag	bags	21
pants	leg	22
shorts	leg	23
skirt	leg	24
white	leg color	25
red	leg color	26
brown	leg color	27
yellow	leg color	28
green	leg color	29
blue	leg color	30
gray	leg color	31
purple	leg color	32
black	leg color	33
shoes	foot	34
sandal	foot	35
hidden	foot	36
no color	foot color	37
white	foot color	38
colorful	foot color	39
black	foot color	40
young	age	41
teenager	age	42
adult	age	43
old	age	44
id	id	45

Prediction Sample

How to use

first download Market-1501.

from delivery import data_delivery

main_path = './datasets/Market1501/Market-1501-v15.09.15/gt_bbox/'
path_attr = './attributes/CA_Market_with_id.npy'

attr = data_delivery(main_path = main_path,
              path_attr = path_attr,
              need_id = True,
              need_parts = True,
              need_attr = False)

The output of data_delivery would be a python dictionary. If you set need_id True, it will include ids. If set need_parts True, it will include 11 key and value which are attributes for each part of dataset. If set need_attr = True, it will include a key 'attributes', a vector with size of (25259, 45) which is all attributes together.

Citation

If you use this dataset in your research, please kindly cite our work as,

@article{
CA-Market: A Partially Categorical AnnotatingApproach Based on Market1501 Dataset for Attribute Detection
}

CA_Market1501 Dataset:

@inproceedings{bodaghi2021market,
  title={CA-Market: A Partially Categorical AnnotatingApproach Based on Market1501 Dataset for Attribute Detection},
  author={Bodaghi, Hossein and Samiee, Shayan and Masoulehe, Mehdi Tale and Kalhor, Ahmad},
  booktitle={2021 7th International Conference on Signal Processing and Intelligent Systems (ICSPIS)},
  pages={01--08},
  year={2021},
  organization={IEEE}
}