search

DART2022 / DART

DART: Articulated Hand Model with Diverse Accessories and Rich Textures (NeurIPS 2022 - Datasets and Benchmarks Track)
https://dart2022.github.io/
132 stars 9 forks source link

Fine-grained classification tags #10

Closed walsvid closed 1 year ago

walsvid commented 1 year ago

Hi authors, thank you for your great work.

I wonder if DARTset data has annotation files for finer grained semantic description of each sample. For example, whether the hand corresponding to a certain picture has accessories, what kind of skin color and other semantic labels.

I notice that the pkl file contains ['pose', 'joint3d', 'vertex', 'joint2d', 'img'], but no other tags.

For example, as it in the "Ablation Study On Accessories" in the paper, you have report the statistics on w/o. Acs and w/ O. Acs and their respective performance improvements.

I wonder how to select sub-dataset with different attributes quickly?

tomguluson92 commented 1 year ago

Hi @walsvid , thanks for your interests. The ablation study we conduct is on mano pose sampled from Freihand and do not publicly included in DARTset. For you interest, I guess you could generate your own data from the Unity code and may be you need to revise it a little bit.