Figure 1: Examples from our test set of motions. The left two images are natural (motion capture data). The two images to the right are unnatural (badly edited and incompletely cleaned motion). Joints that are marked in red-yellow were detected as having unnatural motion. Frames for these images were selected by the method presented in [Assa et al. 2005]. In this paper, we investigate whether it is possible to develop a measure that quantifies the naturalness of human motion (as defined by a large database). Such a measure might prove useful in verifying that a motion editing operation had not destroyed the naturalness of a motion capture clip or that a synthetic motion transition was within the space of those seen in natural human motion. We explore the performance of mixture of Gaussians (MoG), hidden Markov models (HMM), and switching linear dynamic systems (SLDS) on this problem. We use each of these statistical models alone and as part of an ensemble of smaller statistical models. We also implement a Naive Bayes (NB) model for a baseline comparison. We test these techniques on motion capture data held out from a database, keyframed motions, edited motions, motions with noise added, and synthetic motion transitions. We present the results as receiver operating characteristic (ROC) curves and compare the results to the judgments made by subjects in a user study.

To my wife Wei and my son Billy. iv Generating human motion that appears natural is a long standing problem in character animation. Researchers have explored many different approaches including physics-based simulation, optimization, and data-driven methods such as motion graphs and motion interpolation. One major difficulty in applying most of these approaches is the lack of an implementable definition of what it means for motion to be natural or human-like. In this thesis, we explore two techniques to fill this gap. The first technique creates a naturalness measure for quantifying natural human motion. The second technique involves a statistical analysis of human motion to compute aggregate statistics that are needed to guide animation algorithms for human figures toward natural looking solutions. A naturalness measure should be useful in verifying that a motion editing

www.ppgia.pucpr.br Abstract. A novel approach for the detection and classification of human movements in videos scenes is presented in this paper. It consists in detecting, segmenting and tracking foreground objects in video scenes to further classify their movements as conventional or non-conventional. From each tracked object in the scene, features such as position, speed, changes in direction and temporal consistency of the bounding box dimension are extracted. These features make up feature vectors that are stored together with labels that categorize the movement and which are assigned by human supervisors. At the classification step, an instancebased learning algorithm is used to classify the object movement as conventional or non-conventional. For this aim, feature vectors computed from objects in motion are matched against reference feature vectors previously labeled. Experimental results on video clips from two different databases (Parking Lot and CAVIAR) have shown that the proposed approach is able to detect non-conventional human movements in video scenes with accuracies between 77 % and 82%. Keywords: HumanMovement Classification, ComputerVision, Security. 1

Abstract — This article presents a novel approach for detection of non-conventional events in videos scenes. This novel approach consists in analyzing in real-time video from a security camera to detect, segment and tracking objects in movement to further classify its movement as conventional or non-conventional. From each tracked object in the scene features such as position, speed, changes in directions and in the bounding box sizes are extracted. These features make up a feature vector. At the classification step, feature vectors generated from objects in movement in the scene are matched almost in real-time against reference feature vectors previously labeled which are stored in a database and an algorithm based on the instance-based learning paradigm is used to classify the object movement as conventional or non-conventional. Experimental results on video clips from two databases (Parking Lot and CAVIAR) have shown that the proposed approach is able to detect non-conventional events with accuracies between 77% and 82%.

Our goal in this project was to implement and validate the results of recent work employing statistical techniques to automatically determine a “naturalness” measure of human motion data. Using a training set of motion capture data (that in effect embodies our definition of naturalness), we learn several models to represent this natural motion, and test them on a variety of hand-selected positive and negative examples. The models we consider are the Naive Bayes model, mixtures of Gaussians, and hidden Markov models. We restrict our study to walking motions, but nonetheless achieve convincing and meaningful results, which are illustrated using ROC curves. In addition, we mention shortcomings of the original paper, and provide a few suggestions for further work.