Hugging Face's logo

Datasets:
zyuzuguldu
/
deepfashion2-upper-body-masks

Tasks:
Image Segmentation
Modalities:
Image
Text
Formats:
parquet
Size:
100K - 1M
Tags:
fashion
deepfashion2
garment-segmentation
upper-body
masks
Libraries:
Datasets
pandas
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
Dataset Viewer
Auto-converted to Parquet Duplicate
image_id
stringlengths
6
6
mask
imagewidth (px)
68
1.14k
width
int32
68
1.14k
height
int32
71
1.84k
mask_ratio
float32
0
1
000001
468
624
0.6019
000002
468
702
0.5021
000009
468
623
0.4716
000010
468
624
0.3394
000011
750
961
0.0234
000012
640
694
0.0682
000013
468
702
0.3049
000014
468
702
0.2897
000015
640
710
0.3837
000016
640
529
0.0603
000017
640
487
0.2916
000018
750
949
0.0325
000019
750
1,125
0.0824
000020
640
545
0.4648
000021
468
623
0.0485
000022
640
407
0.0125
000023
640
420
0.1031
000024
640
401
0.0087
000025
640
419
0.1105
000026
640
674
0.0159
000027
640
653
0.02
000028
640
429
0.023
000029
468
702
0.2457
000030
640
414
0.011
000031
640
415
0.0074
000032
640
671
0.0219
000034
640
960
0.0805
000035
640
960
0.0519
000036
640
960
0.0735
000039
468
624
0.3497
000040
468
579
0.117
000041
468
624
0.0928
000042
468
627
0.3706
000043
468
625
0.1499
000044
468
833
0.0632
000045
468
624
0.4004
000046
468
631
0.1369
000047
468
625
0.1982
000048
468
624
0.3977
000049
468
826
0.4503
000050
468
833
0.5103
000051
468
833
0.228
000052
468
833
0.2923
000053
750
1,000
0.1543
000054
750
1,000
0.1424
000055
750
1,000
0.1539
000056
750
1,000
0.1697
000057
750
1,000
0.1502
000058
750
1,000
0.1508
000059
750
1,000
0.115
000060
750
968
0.1149
000061
750
1,000
0.1277
000062
468
702
0.3002
000063
468
833
0.1256
000064
468
624
0.1279
000065
468
624
0.6785
000066
468
624
0.4488
000067
468
624
0.6871
000068
468
624
0.0493
000069
468
702
0.2286
000070
720
960
0.312
000071
468
832
0.5209
000072
468
831
0.4564
000073
750
520
0.0185
000074
750
1,145
0.0242
000075
750
1,145
0.0276
000076
750
520
0.0181
000077
468
702
0.2349
000078
750
1,145
0.0429
000079
750
1,145
0.0319
000080
750
1,145
0.0197
000081
750
1,145
0.0312
000082
750
520
0.025
000083
750
520
0.0213
000084
468
625
0.0468
000085
374
664
0.7217
000086
468
832
0.771
000087
800
697
0.2406
000088
800
625
0.1076
000089
800
706
0.1511
000090
800
682
0.1442
000091
291
516
0.4652
000092
880
467
0.1364
000093
880
653
0.0542
000094
880
609
0.0601
000095
837
595
0.1929
000096
835
598
0.1825
000097
468
702
0.5002
000098
468
702
0.5061
000099
880
530
0.1569
000100
880
577
0.0545
000101
468
830
0.5016
000102
468
764
0.2237
000103
468
624
0.5403
000104
468
624
0.7125
000105
468
631
0.6748
000106
468
624
0.1586
000107
468
725
0.4003
000108
468
702
0.4473
000109
468
702
0.3111
End of preview. Expand in Data Studio

DeepFashion2 Upper Body Garment Segmentation Masks

📋 Dataset Description

Binary segmentation masks for upper-body garments from DeepFashion2 dataset.

Statistics

Split Masks Size
Train 137,850 ~686 MB
Validation 22,767 ~110 MB
Total 160,617 ~796 MB

🚀 Quick Start 

from datasets import load_dataset

# Load dataset
dataset = load_dataset("zyuzuguldu/deepfashion2-upper-body-masks")

# Access a sample
sample = dataset["train"][0]
mask = sample["mask"]  # PIL Image
image_id = sample["image_id"]

🎯 Use Cases

  • Training segmentation models (U-Net, DeepLab, etc.)
  • Virtual try-on applications
  • Fashion image editing
  • Benchmark evaluation

📝 Dataset Structure

Each sample contains:

  • image_id: Unique identifier
  • mask: Binary mask (PIL Image)
  • width: Mask width
  • height: Mask height
  • mask_ratio: Garment coverage (0.0-1.0)

🔗 Related

⚖️ License

Apache 2.0 - Derived from DeepFashion2 dataset

Made with ❤️ for the fashion-tech community
Downloads last month
23
Number of rows:
160,617
Total file size:
415 MB

Models trained or fine-tuned on zyuzuguldu/deepfashion2-upper-body-masks