Seeing, Sensing and Recognizing Laban Movement Qualities
summary
Argues that Laban Efforts become computationally recognizable when model design follows expert observation and uses multimodal capture: EMG + accelerometer + Vicon to record muscle activation, norm of jerk, and elbow‑to‑chest opening. Trains real‑time hidden Markov models on the eight basic effort actions, finds dynamic features best for time and EMG best for weight while space remains hard to capture, and shows the fused multimodal feature set performs significantly better. System runs live with dancers and supplies immediate feedback; segmentation of phrases is done manually.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 0:20Laban Movement Analysis Framework
- 0:38Why is Movement Important
- 1:14Movement is Complex
- 1:46Dance and Movement
- 3:43Effort
- 5:12Effort Actions
- 6:57Laban Experts
- 7:18Interview
- 9:41Computational System
- 11:54Quality of Recognition
- 12:45Results
- 15:34Contributions
- 16:25Conclusion
- 16:45Questions
related talks
28mDeep Learning-Based Human Activity Recognition for Physical Skills...
26mBiosensor Machine Learning with Julia
28mStrike a Pose - Gesture recognition in JavaScript w/ ML & Arduino
31mData as an instrument for more efficient collaboration with robots
07mMultimodal Emotion Recognition of Hand-Object Interaction
22mStrike a Pose: Gesture Recognition in JS with Machine Learning
28mEnhancing Upper Limb Rehabilitation with Unsupervised Machine Learning Models
31mJennifer Marsman, Principal Developer Evangelist, Microsoft