Overview
We extracted frames from 100DOH videos to build a new 100K frame-level dataset. Primarily, 85% of the frames comes from the subset of 100DOH and the rest 15% comes from
VLOG Dataset (a 3x extended and relabeled version). We randomly picked frames and filtered out frames that do not contain hands.
For each frame, we provide annotations for:
- hand bounding box
- hand side: {left (L), right (R)}
- hand contact state: {no contact (N), self-contact (S), other person contact (O), portable object contact (P), stationary object contact (F, short for furniture)}
- object bounding box that the person is contacting irrespective of class name
- association between hand and in-contact object
Before using the dataset or models derived from it, we want our users to be aware that the dataset may not be entirely representative. The dataset is collected from YouTube via keyword searches in English. It covers diverse interaction activities, but it might not representative of all demographics and due to its sources, it is likely there are underlying biases and spurious correlations (e.g., certain people doing certain activities). Like all data, one should be cautious and skeptical.
The number of annotations for each type:
|
|
#Frames |
#Hand Box |
#Hand Side |
#Contact State |
#Object Box |
Total |
99,899 |
189,426 |
189,426 |
189,426 |
140,431 |
(Initially, it contains 100K frames. However, when we went through the frames, we observed some objectionable ones and we decided to deleted the 101 objectionable ones we found. This dataset is enormous and it is impossible to analyze everything. We have done our best but are committed to making the dataset as good as possible. Please email us with any issues.)
|
Visualization
Explore Hand Size & Contact State
Select a set of hands to look at:
As you mouse around the image, you'll see an image with a hand at that location below:
You can see the prior change as you slide across different percentiles of the dataset according to size. A sample image from that percentile appears
on the right.
Select a hand type:
All
Left
Right