y******u
发帖数: 1460 | 1 Dear fellow researchers,
I would like to introduce a new RGB-Depth human daily activity video
benchmark. It has 12 action categories such as make a phone call, eat meal,
mop the floor etc. It totally has 1189 manually segmented and annotated
video segments. Synchronized RGB and depth images are provided at 30 FPS.
The activities are performed by 30 subjects.
You can now download this video dataset from two sources:
1) Dropbox: you can log into the dropbox using the following public
account and password to download the whole dataset.
Account: a**************[email protected] password: ADSC2013
2) FTP site of ADSC: please follow the instruction given by http://adsc.illinois.edu/sites/default/files/files/ADSC-RGBD-dataset-download-instructions.pdf
to access the data.
Any questions regarding the dataset, please drop an Email to: bingbing.ni@
adsc.com.sg
Reference: Rgbd-hudaact: A color-depth video database for human daily
activity recognition
B Ni, G Wang, P Moulin, Consumer Depth Cameras for Computer Vision, 2013. | g*****g
发帖数: 43 | |
|
