Depth-based hand pose estimation: data, methods, and challenges

James Supancic 1 Gregory Rogez 2 Yi Yang 3 Jamie Shotton 4 Deva Ramanan 5
1 UCI - University of California [Irvine]
2 LEAR - Learning and recognition in vision
Inria Grenoble - Rhône-Alpes, LJK - Laboratoire Jean Kuntzmann, INPG - Institut National Polytechnique de Grenoble
3 Baidu Research
4 Microsoft - Microsoft Research [Cambridge]
5 CMU - Carnegie Mellon University [Pittsburgh]
Abstract : Hand pose estimation has matured rapidly in recent years. The introduction of commodity depth sensors and a multitude of practical applications have spurred new advances. We provide an extensive analysis of the state-of-the-art, focusing on hand pose estimation from a single depth frame. To do so, we have implemented a considerable number of systems, and will release all software and evaluation code. We summarize important conclusions here: (1) Pose estimation appears roughly solved for scenes with isolated hands. However, methods still struggle to analyze cluttered scenes where hands may be interacting with nearby objects and surfaces. To spur further progress we introduce a challenging new dataset with diverse, cluttered scenes. (2) Many methods evaluate themselves with disparate criteria , making comparisons difficult. We define a consistent evaluation criteria, rigorously motivated by human experiments. (3) We introduce a simple nearest-neighbor baseline that outperforms most existing systems. This implies that most systems do not generalize beyond their training sets. This also reinforces the under-appreciated point that training data is as important as the model itself. We conclude with directions for future progress.
Document type :
Conference papers
Complete list of metadatas

Cited literature [37 references]  Display  Hide  Download

screenshot_handEval.jpg
screenshot_handEval.jpg
https://hal.inria.fr/hal-01237023
Contributor : Gregory Rogez <>
Submitted on : Wednesday, December 2, 2015 - 3:29:14 PM
Last modification on : Thursday, August 8, 2019 - 1:10:32 AM
Long-term archiving on: Saturday, April 29, 2017 - 3:45:06 AM

Files

HandEval.pdf
Files produced by the author(s)

Identifiers

Collections

CNRS | INRIA | LJK | UGA | LJK_GI | LJK_GI_LEAR

Citation

James Supancic, Gregory Rogez, Yi Yang, Jamie Shotton, Deva Ramanan. Depth-based hand pose estimation: data, methods, and challenges. ICCV - IEEE International Conference on Computer Vision, Dec 2015, Santiago, Chile. pp.1868-1876, ⟨10.1109/ICCV.2015.217⟩. ⟨hal-01237023⟩

Export

BibTeX TEI DC DCterms EndNote

Share

Metrics

Record views

9209

Files downloads

576

 

 

Contact

archive-ouverte@inria.fr