Distances evolution analysis for online and off-line human object interaction recognition

Abstract

Human action recognition in 3D sequences is one of the most challenging and active areas of research in the computer vision domain. However designing automatic systems that are robust to significant variability due to object combinations and high complexity of human motions are more challenging in addition to the typical requirements such as rotation, translation, and scale invariance is challenging task. In this paper, we propose a spatio-temporal modeling of human-object interaction videos for on-line and off-line recognition. The inter joint distances and the object are considered as low-level features for online classification. For off-line recognition, we propose rate-invariant classification of full video and early recognition. A shape analysis of trajectories of the inter-joint and object-joints distances is proposed for this end. The experiments conducted following state-of-the-art settings using MSR Daily Activity 3D Dataset and Online RGBD Action Dataset and on a new Multi-view dataset for human object interaction demonstrate that the proposed approach is effective and discriminative for human object interaction classification as demonstrated here.

A Comprehensive Survey of Vision-Based Human Action Recognition Methods

“…Many studies have considered human action recognition. According to the complexity of the human action, previous techniques can be categorized into the following four classes of action semantics [1] from low to high: primitive action recognition [52], single-person action recognition [6,9], interaction recognition [26], and group action recognition [53]. Primitive action refers to basic movement changes of human body parts, which is the minimum structure of action.…”

“…Action detection refers to locating the position at which an action occurs in time and space from image sequence data that have not been segmented. In recent years, interaction [ 26 , 27 , 28 ] and human action detection [ 29 , 30 , 31 ] have become prominent research topics. However, there is no overarching summary of the methods applicable to these two issues.…”

A Survey of Vision-Based Human Action Evaluation Methods

“…Human action evaluation (HAE) [1,2,3,4,5,6,7,8,9,10,48,49,50] aims to design computation models for representing the dynamic changing process of human movement and to develop evaluation techniques to measure the completion quality of human action. The target of action evaluation not only requires the recognition of the action performed but also more significantly, needs to provide a quality assessment of how the action was performed.…”

“…In recent years, human action evaluation has emerged as an important problem in a variety of computer vision applications, which range from sports training [1,2,3,4,5] to healthcare and physical rehabilitation [6,7,8,9], interactive entertainment [10,11,12], and video understanding [13,14,15]. In contrast to the aims of traditional human action recognition to infer the class label from predefined action categories (action classification [16,17]), to locate the starting and end positions of actions (action detection [18,19]), and to predict the future state of actions on the basis of incomplete action observations (action prediction [20,21]), the target of human action evaluation is to make computers automatically quantify how well people perform actions and further provide interpretable feedback for improving the motion of the human body.…”

Learning Effective Skeletal Representations on RGB Video for Fine-Grained Human Action Quality Assessment

“…Human action evaluation (HAE) aims to tackle the challenging problem of making computers automatically quantify how well people perform actions. It has been largely unexplored in past decades [1,2], and has been involved in a wide range of applications, such as sport activity scoring and training systems [1,3], physical therapy and rehabilitation [4][5][6][7], interactive entertainment [8][9][10], skill training for expertise learners, and video retrieval [11][12][13]. With the rapid progress in human activity understanding in the research area of computer vision, research efforts have recently been devoted to human action quality assessment [14,15].…”