Assessing similarity models for human‐motion retrieval applications

Abstract

The development of motion capturing devices poses new challenges in the exploitation of human-motion data for various application fields, such as computer animation, visual surveillance, sports, or physical medicine. Recently, a number of approaches dealing with motion data have been proposed, suggesting characteristic motion features to be extracted and compared on the basis of similarity. Unfortunately, almost each approach defines its own set of motion features and comparison methods; thus, it is hard to fairly decide which similarity model is the most suitable for a given kind of human-motion retrieval application. To cope with this problem, we propose the human motion model evaluator, which is a generic framework for assessing candidate similarity models with respect to the purpose of the target application. The application purpose is specified by a user in form of a representative sample of categorized motion data. Respecting such categorization, the similarity models are assessed from the effectiveness and efficiency points of view using a set of space-complexity, information-retrieval, and performance measures. The usability of the framework is demonstrated by case studies of three practical examples of retrieval applications focusing on recognition of actions, detection of similar events, and identification of subjects.

ExpressGesture: Expressive gesture generation from speech through database matching

“…(7) Synthetic preparation and retraction phases are generated to link the gestures in the sequence combining motion matching with neural network controllers, including compression of motion data to a low-dimensional representation, memory usage was reduced significantly by Holden et al 19 A number of other works have proposed methods for motion database retrieval based on sets of motion parameter keys. 20,21 Relating to both statistical and machine-learning models, our method aims to combine the power of neural networks to capture the relationship of speech features with higher level motion features with the advantage of realistic gesture motion, sampled from a large database.…”

Precision strength training: Data-driven artificial intelligence approach to strength and conditioning

“…The technique assessment models (or motion similarity models [726,835]), are the most relevant from these models for our precision strength training assessment for kinematic assessment of the athlete. In powerlifting [225,320,821], the goal is to produce maximum force, in a relatively long time window [829], in the three powerlifting lifts: squat [154,390], bench press [605,876] and deadlift [217,390].…”

Retrieval of spatial–temporal motion topics from 3D skeleton data

“…Kapadia et al [13] introduce the concept of motion keys which facilitates subsequence matching of large datasets. Valcik et al [35] use the alternatives of dynamic time warping (DTW) to align motion clips and keep salient spatial properties at the same time. Lee et al [20] develop a multi-term LSTM module to understand temporal dynamics of motion by adjusting the length and stride of a sliding window.…”