A scheme is developed for classifying the types of motion perceived by a humanlike robot. It is assumed that the robot receives visual images of the scene using a perspective system model. Equations, theorems, concepts, clues, etc., relating the objects, their positions, and their motion to their images on the focal plane are presented. >

http://ieeexplore.ieee.org/iel2/815/3148/00100651.pdf

Robot vision

Human actions can be represented by the trajectories of skeleton joints. Traditional methods generally model the spatial structure and temporal dynamics of human skeleton with hand-crafted features and recognize human actions by well-designed classifiers. In this paper, considering that recurrent neural network (RNN) can model the long-term contextual information of temporal sequences well, we propose an end-to-end hierarchical RNN for skeleton based action recognition. Instead of taking the whole skeleton as the input, we divide the human skeleton into five parts according to human physical structure, and then separately feed them to five subnets. As the number of layers increases, the representations extracted by the subnets are hierarchically fused to be the inputs of higher layers. The final representations of the skeleton sequences are fed into a single-layer perceptron, and the temporally accumulated output of the perceptron is the final decision. We compare with five other deep RNN architectures derived from our model to verify the effectiveness of the proposed network, and also compare with several other methods on three publicly available datasets. Experimental results demonstrate that our model achieves the state-of-the-art performance with high computational efficiency.

/pdf/hierarchical-recurrent-neural-network-for-skeleton-based-18vsz9c292.pdf

Hierarchical recurrent neural network for skeleton based action recognition

Feature selection is an important task in data mining and machine learning to reduce the dimensionality of the data and increase the performance of an algorithm, such as a classification algorithm. However, feature selection is a challenging task due mainly to the large search space. A variety of methods have been applied to solve feature selection problems, where evolutionary computation (EC) techniques have recently gained much attention and shown some success. However, there are no comprehensive guidelines on the strengths and weaknesses of alternative approaches. This leads to a disjointed and fragmented field with ultimately lost opportunities for improving performance and successful applications. This paper presents a comprehensive survey of the state-of-the-art work on EC for feature selection, which identifies the contributions of these different algorithms. In addition, current issues and challenges are also discussed to identify promising areas for future research.

/pdf/a-survey-on-evolutionary-computation-approaches-to-feature-5c2qznm5vr.pdf

A Survey on Evolutionary Computation Approaches to Feature Selection

Going deeper into action recognition

There is sufficient evidence proving the impact that negative lifestyle choices have on people’s health and wellness. Changing unhealthy behaviours requires raising people’s self-awareness and also providing healthcare experts with a thorough and continuous description of the user’s conduct. Several monitoring techniques have been proposed in the past to track users’ behaviour; however, these approaches are either subjective and prone to misreporting, such as questionnaires, or only focus on a specific component of context, such as activity counters. This work presents an innovative multimodal context mining framework to inspect and infer human behaviour in a more holistic fashion. The proposed approach extends beyond the state-of-the-art, since it not only explores a sole type of context, but also combines diverse levels of context in an integral manner. Namely, low-level contexts, including activities, emotions and locations, are identified from heterogeneous sensory data through machine learning techniques. Low-level contexts are combined using ontological mechanisms to derive a more abstract representation of the user’s context, here referred to as high-level context. An initial implementation of the proposed framework supporting real-time context identification is also presented. The developed system is evaluated for various realistic scenarios making use of a novel multimodal context open dataset and data on-the-go, demonstrating prominent context-aware capabilities at both low and high levels.

/pdf/human-behavior-analysis-by-means-of-multimodal-context-4x6dd2ny19.pdf

Human Behavior Analysis by Means of Multimodal Context Mining.

Human Behaviour Analysis (HBA) is more and more being of interest for computer vision and artificial intelligence researchers. Its main application areas, like Video Surveillance and Ambient-Assisted Living (AAL), have been in great demand in recent years. This paper provides a review on HBA for AAL and ageing in place purposes focusing specially on vision techniques. First, a clearly defined taxonomy is presented in order to classify the reviewed works, which are consequently presented following a bottom-up abstraction and complexity order. At the motion level, pose and gaze estimation as well as basic human movement recognition are covered. Next, the mainly used action and activity recognition approaches are presented with examples of recent research works. Increasing the degree of semantics and the time interval involved in the HBA, finally the behaviour level is reached. Furthermore, useful tools and datasets are analysed in order to provide help for initiating projects.

https://rua.ua.es/dspace/bitstream/10045/33749/1/2012_Chaaraoui_etal_ESA-pre.pdf

A review on vision techniques applied to Human Behaviour Analysis for Ambient-Assisted Living

https://rua.ua.es/dspace/bitstream/10045/33750/1/2013_Chaaraoui_etal_ESA-pre.pdf

Silhouette-based human action recognition using sequences of key poses

This work is the first comprehensive review about visual privacy protection.Several categories are established for all of the available privacy protection methods.A review of intelligent monitoring systems that takes privacy into account is provided.An overall vision of the state of the art is given to introduce novel researchers.A critical discussion about relevant key topics and open issues is provided. Recent advances in computer vision technologies have made possible the development of intelligent monitoring systems for video surveillance and ambient-assisted living. By using this technology, these systems are able to automatically interpret visual data from the environment and perform tasks that would have been unthinkable years ago. These achievements represent a radical improvement but they also suppose a new threat to individual's privacy. The new capabilities of such systems give them the ability to collect and index a huge amount of private information about each individual. Next-generation systems have to solve this issue in order to obtain the users' acceptance. Therefore, there is a need for mechanisms or tools to protect and preserve people's privacy. This paper seeks to clarify how privacy can be protected in imagery data, so as a main contribution a comprehensive classification of the protection methods for visual privacy as well as an up-to-date review of them are provided. A survey of the existing privacy-aware intelligent monitoring systems and a valuable discussion of important aspects of visual privacy are also provided.

https://eprints.kingston.ac.uk/31628/1/Florez-Revuelta-F-31628.pdf

Visual privacy protection methods

Interest in RGB-D devices is increasing due to their low price and the wide range of possible applications that come along. These devices provide a marker-less body pose estimation by means of skeletal data consisting of 3D positions of body joints. These can be further used for pose, gesture or action recognition. In this work, an evolutionary algorithm is used to determine the optimal subset of skeleton joints, taking into account the topological structure of the skeleton, in order to improve the final success rate. The proposed method has been validated using a state-of-the-art RGB action recognition approach, and applying it to the MSR-Action3D dataset. Results show that the proposed algorithm is able to significantly improve the initial recognition rate and to yield similar or better success rates than the state-of-the-art methods.

https://rua.ua.es/dspace/bitstream/10045/33751/1/2013_Chaaraoui_etal_ESA-pre.pdf

Evolutionary joint selection to improve human action recognition with RGB-D devices

Since the Microsoft Kinect has been released, the usage of marker-less body pose estimation has been enormously eased. Based on 3D skeletal pose information, complex human gestures and actions can be recognised in real time. However, due to errors in tracking or occlusions, the obtained information can be noisy. Since the RGB-D data is available, the 3D or 2D shape of the person can be used instead. However, depending on the viewpoint and the action to recognise, it might present a low discriminative value. In this paper, the combination of body pose estimation and 2D shape, in order to provide additional characteristic value, is considered so as to improve human action recognition. Using efficient feature extraction techniques, skeletal and silhouette-based features are obtained which are low dimensional and can be obtained in real time. These two features are then combined by means of feature fusion. The proposed approach is validated using a state-of-the-art learning method and the MSR Action3D dataset as benchmark. The obtained results show that the fused feature achieves to improve the recognition rates, outperforming state-of-the-art results in recognition rate and robustness.

/pdf/fusion-of-skeletal-and-silhouette-based-features-for-human-39opoq9utn.pdf

Alexandros Andre Chaaraoui

Papers

A review on vision techniques applied to Human Behaviour Analysis for Ambient-Assisted Living

Silhouette-based human action recognition using sequences of key poses

Visual privacy protection methods

Evolutionary joint selection to improve human action recognition with RGB-D devices

Fusion of Skeletal and Silhouette-Based Features for Human Action Recognition with RGB-D Devices