Official repository for CVPR2021: UAV-Human: A Large Benchmark for Human Behavior Understanding with Unmanned Aerial Vehicles
[CVF OpenAccess] [arXiv] [ResearchGate]
The dataset is available for Download now!
VideoNames: P070S01G10B00H10UC102000LC092000A031R0_09131758.avi
P070: (PersonID) unique person ID for the main subject in current video
S01: (SetupID) setup id that indicates changes of clothing, hat, backpack of the main subject
G10: (Gender) first bit represents main subject's gender, second bit represents auxiliary subject's gender
0: n/a; 1: male; 2: female
B00: (Backpack) first bit represents main subject's backpack color, second bit represents auxiliary subject's backpack color
0: n/a; 1: red; 2: black; 3: green; 4: yellow
H10: (Hat) first bit represents main subject's hat color, second bit represents auxiliary subject's hat color
0: n/a; 1: red; 2: black; 3: green; 4: white
UC102000: (UpperClothing) first three bits represent main subject's upper clothing color (2 bits) and style (1 bit), last three bits represent auxiliary subject's upper clothing color (2 bits) and style (1 bit)
color: 0: n/a; 1: red; 2: black; 3: blue; 4: green; 5: multicolor; 6: grey; 7: white; 8: yellow; 9: dark brown; 10: purple; 11: pink
style: 0: n/a; 1: long; 2: short; 3: skirt
LC102000: (LowerClothing) first three bits are main subject's lower clothing color (2 bits) and style (1 bit), last three bits are auxiliary subject's lower clothing color (2 bits) and style (1 bit)
color: 0: n/a; 1: red; 2: black; 3: blue; 4: green; 5: multicolor; 6: grey; 7: white; 8: yellow; 9: dark brown; 10: purple; 11: pink
style: 0: n/a; 1: long; 2: short; 3: skirt
A031: (Action) action labels of current sample
R00: (Replicate) replicate capturing
09131758: capturing timestamp, month(2 bits)/day(2 bits)/hour(2 bits)/minute(2 bits)
In cross-subject-v1 evaluation, we split 119 subjects into training and testing groups. The IDs of training subjects are 0, 2, 5, 6, 7, 8, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 25, 26, 27, 28, 29, 30, 32, 33, 34, 35, 36, 37, 38, 39, 40, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 55, 56, 57, 59, 61, 62, 63, 64, 65, 67, 68, 69, 70, 71, 73, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 98, 100, 102, 103, 105, 106, 110, 111, 112, 114, 115, 116, 117, 118; the remaining subjects are for testing.
In cross-subject-v2 evaluation, we split 119 subjects into training and testing groups. The IDs of training subjects are 0, 3, 4, 5, 6, 8, 10, 11, 12, 14, 16, 18, 19, 20, 21, 22, 24, 26, 29, 30, 31, 32, 35, 36, 37, 38, 39, 40, 43, 44, 45, 46, 47, 49, 52, 54, 56, 57, 59, 60, 61, 62, 63, 64, 66, 67, 69, 70, 71, 72, 73, 74, 75, 77, 78, 79, 80, 81, 83, 84, 86, 87, 88, 89, 91, 92, 93, 94, 95, 96, 97, 99, 100, 101, 102, 103, 104, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 117, 118; the remaining subjects are for testing.
Modality | CSv1 - Acc (%) | CSv2 - Acc (%) |
---|---|---|
RGB Video | 23.86 | 29.53 |
Fisheye Video | 20.76 | 34.12 |
Methods | CSv1 - Acc (%) | CSv2 - Acc (%) |
---|---|---|
DGNN | 29.90 | - |
ST-GCN | 30.25 | 56.14 |
2s-AGCN | 34.84 | 66.68 |
HARD-Net | 36.97 | - |
Shift-GCN | 37.98 | 67.04 |
Please refer to utils/convert_videos_to_frames.py and following:
python convert_videos_to_frames.py --videos path/to/all/videos --frames path/to/output/frames
@InProceedings{Li_2021_CVPR,
author = {Li, Tianjiao and Liu, Jun and Zhang, Wei and Ni, Yun and Wang, Wenqian and Li, Zhiheng},
title = {UAV-Human: A Large Benchmark for Human Behavior Understanding With Unmanned Aerial Vehicles},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2021},
pages = {16266-16275}
}
tianjiao_li [at] mymail.sutd.edu.sg