Toward Realistic Hands Gesture Interface: Keeping it Simple for Developers and Machines
summary
Introduces a developer-facing gesture SDK that uses Intel RealSense SR300 depth cameras and a simple pose language built from six-direction finger/palm predicates (direction, relative location, touching, folding) so gestures are files or GUI-defined state machines. Recognition runs with a pipeline of fast classifiers achieving about 10 ms per frame and similar accuracy to prior work on the NYU dataset while being an order of magnitude faster on CPU. Gesture detection hits near 100% for many gestures, degrades past ~60 cm, and needs ~1 minute of user tuning for ~80% first-time success.
AI-written summary · about
chapters (7)
- 0:00Introduction and team
- 1:36Challenges with articulated skeletons
- 4:51Project Prague approach
- 8:47Technology behind recognition
- 12:34Live gesture demo
- 13:18Killer applications and scenarios
- 14:02Q&A session
related talks
23mCreate Amazing Experiences Using Intel® RealSense Technology
14mBack-Hand-Pose: 3D Hand Pose Estimation for a Wrist-worn Camera via Dorsum Deformation Network
07mFingerTrak: Continuous 3D Hand Pose Tracking by Deep Learning Hand Silhouettes Captured by ...
05mTouchPose: Hand Pose Prediction, Depth Estimation, and Touch Classification from Capacitive Images
21mPosePIE: Replace Your Keyboard and Mouse With AI-Driven Gesture Control
14mDeepFisheye: Near-Surface Multi-Finger Tracking Technology Using Fisheye Camera
10mProactive Sensing for Improving Hand Pose Estimation
08mGesturAR: An Authoring System for Creating Freehand Interactive Augmented Reality Applications