The M3D labelling system for gesture annotation has recently developed a valuable website with training materials and videos guiding participants through each step of the gesture annotation process via several challenges:
https://m3d.upf.edu/ (the M3D training website)
The M3D Training Program contains training material destined to anyone interested in the multimodal labeling of video corpora. It is based on the manual for The MultiModal MultiDimensional (M3D) labeling system, developed by the Prosodic Studies Group, Universitat Pompeu Fabra, Barcelona; the Faculty of Psychology and Education Sciences, Universitat Oberta de Catalunya, Barcelona and the Speech Communication Group at the Research Laboratory of Electronics, Massachusetts Institute of Technology, USA.
The goal of this training program is to introduce annotators to the basic multimodal labeling scheme of M3D. The labeling system adopts a multidimensional approach to gesture analysis which takes into account the form dimension, the prosodic dimension, and the meaning dimension of gestures. This training program is designed for anyone interested in studying gestures, and will demonstrate how to work with M3D. (from the M3D website)