Discuss about the Project Management for Human Behavior and Activity.
Human behavior and activity patterns can be modeled for detection or recognition of the special events, which has been the case of interest for various researchers in recent years (Mubashir, Shao and Seed 2013). The diverse methods are implicated in order to build intelligent vision system. The system aims at making correct semantic inferences and understanding form the observed dynamics of moving targets. Most of the applications are in video context retrieval, surveillance and human computer interfaces. The paper focuses on reviewing the literature as proposed by various researchers related to the human behavior detection especially through the computer vision. The main objective of the paper is to characterize the literature such that to bring the key challenges to attention.
Human Behavior analysis and understanding (HBA/U) is study, which involves a variety of investigation fields rendering from the implementation of the high-level abstraction behavior models to the motion detection system. According to Chaaraoui, Climent-Pérez and Flórez-Revuelta (2012), the different taxonomies related to the Human behavior analysis are the primary areas of discussion which should be studied in order to get a clear understanding of the human behavior. The authors defined the action taxonomy of humans to be classified into three level of abstraction. First is the action or motion recognition, which is derived from the motor primitives, which represent that entities out of which the actions are built. Second is the set of action primitives, which make up an action. Last is the actual activity, which is involved with a larger scale of events while interaction with the environment or with objects. In this way, a set of multiple actions can be classified at the activity level, which could be helpful in understanding the behavior of humans in a period ranging from the tenth of a second to a minute on the computer-aided vision.
Usually, each human behavior type including the walking running and stopping has a unique property with unique trajectory pattern. According to Xiao et al. (2013), an unusual behavior is the activity being conducted by a suspicious individual who are of doing things is not normal. The unusual behavior is often difficult to define and can be detected through the change in a particular trajectory in respective of a normal behavior. Based on the movement and trajectory other of the subjects, it becomes easy to detect the human behavior to be normal or unusual. This detection of behavior can be aided by the use of the computer vision and video surveillance as well.
According to Chaaraoui, Climent-Pérez and Flórez-Revuelta (2012), this is the most important step for detection of motion where the basic actions is understood by a series of motions either as a whole or a part of the body of the subject. The actions are detectable owing to the different body poses, which are involved and are varied through a short period. According to Popoola and Wang (2012), it is very important to understand the difference between the activity and an action. In this respect, the time lapse and the people involved should be taken into account. As for example, an individual manipulating an object is performing an action while several such actions being performed by the same individual is called an activity.
The computer vision and artificial intelligence researchers have been conducting researches and are interested in the Human behavior Analysis in recent years. According to Chaaraoui, Climent-Pérez and Flórez-Revuelta (2012), the main application areas of the computer vision includes the AAL (Ambient Assisted Living) and the Video Surveillance. The technique includes the deigning of the taxonomy at the initial stage. The next step involves the estimation of the basic human movement at the motion level. Next, the presentations for activity recognition approaches are to be made in respective of the human behavior. At last, the technique requires implementing appropriate tools and datasets in order to analyse the Human Behavior.
According to Cristani et al. (2013), the human behavior can be represented by a continuous video which is segmented into a single behavior pattern. The pattern may consist of interactive activity or a single object where any instant image frame would be representing a class of behavior, which is being visually captured in the form of a video. According to Borges, Conci and Cavallaro (2013), each of the behavior patterns being recorded in the video belongs to the same behavior class. However, the patterns can exhibit considerable variations visually. These characteristics must be considered during the designing of the behavior modeling approach. According to Chaquet, Carmona and Fernández-Caballero (2013), a number of approaches could be adopted for addressing the problems which could be dependent on the nature of the sequence of video being processed.
Methods using full 3D modeling
Most of the researches related to the tracking of the change detection in the human behavior are based on the 3D modeling visual surveillance using computer-generated graphics. The modeling technique requires the use of cameras, which is required to be stabilized. According to Xiao et al. (2013), this modeling technique requires the implementation of Bayesian framework which combines the compact object promotion, shape and appearance as well. Most successful system employs the multiple viewpoints, heavy computation, and good image resolution for the video surveillance. According to Chaquet, Carmona and Fernández-Caballero (2013), the motion recognition can be best pioneered by using motion captured data.
Methods using 2D appearance model
According to Borges, Conci and Cavallaro (2013), the tracking algorithms related to the 2D modeling technique can be classified into two categories including the stochastic method and the deterministic model. The stochastic method uses the space of the state to model the dynamics of the tracking system. While, the deterministic method tracks the motion by performing a search which is iterative for the local maxima of function between the current image and the template image. In addition, there is also an invention of a new system, which is known as the evolutionary technique. According to Xiao et al. (2013), the evolutionary technique helps in extracting the moving targets from the real time video stream. The technique helps in tracking the video stream and classifying the same into predefined categories based on the activity properties.
Gaps in the literature
The literature deals in deifying the taxonomy of human behaviors based on actions. Although, the literature has clearly defined the taxonomy, the higher level approaches including the behavior analysis could be difficult to adapt with the research goals. The literature also lacks the illustration of the frameworks and models related to the detection of the human behavior. The literature also lacks in defining the Human behavior, which could help in differentiating the normal with the unusual behavior of the humans. Moreover, it also lacks in defining the types of usual and unusual behavior such that to allow the establishment of discrimination which could have helped in improving the performance of classifying the human behavior by various computation techniques.
The analysis of the human activities has been one of the most important and intriguing open issues for the automated video surveillance. The activities are detected by the use of computer vision and pattern recognition. The broad range of techniques being illustrated in the paper could be used in the computer vision based on the behavior of humans. The paper helps in organizing the corresponding literature, defining key terms, and discussing the links among the fundamental building blocks, which arise from the action and interaction recognition to the human behavior detection. The paper also helps in providing an illustration of the key aspects of the understanding of human behavior based on the video surveillance.
Borges, P.V.K., Conci, N. and Cavallaro, A., 2013. Video-based human behavior understanding: a survey. IEEE Transactions on Circuits and Systems for Video Technology, 23(11), pp.1993-2008.
Chaaraoui, A.A., Climent-Pérez, P. and Flórez-Revuelta, F., 2012. A review on vision techniques applied to human behaviour analysis for ambient-assisted living. Expert Systems with Applications, 39(12), pp.10873-10888.
Chaquet, J.M., Carmona, E.J. and Fernández-Caballero, A., 2013. A survey of video datasets for human action and activity recognition. Computer Vision and Image Understanding, 117(6), pp.633-659.
Cristani, M., Raghavendra, R., Del Bue, A. and Murino, V., 2013. Human behavior analysis in video surveillance: A social signal processing perspective. Neurocomputing, 100, pp.86-97.
Mubashir, M., Shao, L. and Seed, L., 2013. A survey on fall detection: Principles and approaches. Neurocomputing, 100, pp.144-152.
Popoola, O.P. and Wang, K., 2012. Video-based abnormal human behavior recognition—a review. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 42(6), pp.865-878.
Xiao, Y., Zhu, S., Luo, W., Liu, W. and Huang, D., 2013. Abnormal Behaviour Detection Based on Manifold Learning. Advances in Information Sciences and Service Sciences, 5(3), p.406.