Search results

Social signal processing under affective computing aims at recognizing and extracting useful human social interaction patterns. Fight is a common social interaction in real life. A fight detection system finds wide applications. This paper aims to detect fights in a natural and low-cost manner.

Research works on fight detection are often based on visual features, demanding substantive computation and good video quality. In this paper, the authors propose an approach to detect fight events through motion analysis. Most existing works evaluated their algorithms on public data sets manifesting simulated fights, where the fights are acted out by actors. To evaluate real fights, the authors collected videos involving real fights to form a data set. Based on the two types of data sets, the authors evaluated the performance of their motion signal analysis algorithm, which was then compared with the state-of-the-art approach based on MoSIFT descriptors with Bag-of-Words mechanism, and basic motion signal analysis with Bag-of-Words.

The experimental results indicate that the proposed approach accurately detects fights in real scenarios and performs better than the MoSIFT approach.

By collecting and annotating real surveillance videos containing real fight events and augmenting with well-known data sets, the authors proposed, implemented and evaluated a low computation approach, comparing it with the state-of-the-art approach. The authors uncovered some fundamental differences between real and simulated fights and initiated a new study in discriminating real against simulated fight events, with very good performance.

This paper aims to concentrate on recent innovations in flexible wearable sensor technology tailored for monitoring vital signals within the contexts of wearable sensors and systems developed specifically for monitoring health and fitness metrics.

In recent decades, wearable sensors for monitoring vital signals in sports and health have advanced greatly. Vital signals include electrocardiogram, electroencephalogram, electromyography, inertial data, body motions, cardiac rate and bodily fluids like blood and sweating, making them a good choice for sensing devices.

This report reviewed reputable journal articles on wearable sensors for vital signal monitoring, focusing on multimode and integrated multi-dimensional capabilities like structure, accuracy and nature of the devices, which may offer a more versatile and comprehensive solution.

The paper provides essential information on the present obstacles and challenges in this domain and provide a glimpse into the future directions of wearable sensors for the detection of these crucial signals. Importantly, it is evident that the integration of modern fabricating techniques, stretchable electronic devices, the Internet of Things and the application of artificial intelligence algorithms has significantly improved the capacity to efficiently monitor and leverage these signals for human health monitoring, including disease prediction.

Concentration is the key to safer driving. Ideally, drivers should focus mainly on front views and side mirrors. Typical distractions are eating, drinking, cell phone use, using and searching things in car as well as looking at something outside the car. In this paper, distracted driving detection algorithm is targeting on nine scenarios nodding, head shaking, moving the head 45° to upper left and back to position, moving the head 45° to lower left and back to position, moving the head 45° to upper right and back to position, moving the head 45° to lower right and back to position, moving the head upward and back to position, head dropping down and blinking as fundamental elements for distracted events. The purpose of this paper is preliminary study these scenarios for the ideal distraction detection, the exact type of distraction.

The system consists of distraction detection module that processes video stream and compute motion coefficient to reinforce identification of distraction conditions of drivers. Motion coefficient of the video frames is computed which follows by the spike detection via statistical filtering.

The accuracy of head motion analyzer is given as 98.6 percent. With such satisfactory result, it is concluded that the distraction detection using light computation power algorithm is an appropriate direction and further work could be devoted on more scenarios as well as background light intensity and resolution of video frames.

The system aimed at detecting the distraction of the public transport driver. By providing instant response and timely warning, it can lower the road traffic accidents and casualties due to poor physical conditions. A low latency and lightweight head motion detector has been developed for online driver awareness monitoring.

Existing Simultaneous Localization and Mapping (SLAM) algorithms have been relatively well developed. However, when in complex dynamic environments, the movement of the dynamic points on the dynamic objects in the image in the mapping can have an impact on the observation of the system, and thus there will be biases and errors in the position estimation and the creation of map points. The aim of this paper is to achieve more accurate accuracy in SLAM algorithms compared to traditional methods through semantic approaches.

In this paper, the semantic segmentation of dynamic objects is realized based on U-Net semantic segmentation network, followed by motion consistency detection through motion detection method to determine whether the segmented objects are moving in the current scene or not, and combined with the motion compensation method to eliminate dynamic points and compensate for the current local image, so as to make the system robust.

Experiments comparing the effect of detecting dynamic points and removing outliers are conducted on a dynamic data set of Technische Universität München, and the results show that the absolute trajectory accuracy of this paper's method is significantly improved compared with ORB-SLAM3 and DS-SLAM.

In this paper, in the semantic segmentation network part, the segmentation mask is combined with the method of dynamic point detection, elimination and compensation, which reduces the influence of dynamic objects, thus effectively improving the accuracy of localization in dynamic environments.

Biometric systems are widely used for face recognition. They have rapidly developed in recent years. Compared with other approaches, such as fingerprint recognition, handwriting verification and retinal and iris scanning, face recognition is more straightforward, user friendly and extensively used. The aforementioned approaches, including face recognition, are vulnerable to malicious attacks by impostors; in such cases, face liveness detection comes in handy to ensure both accuracy and robustness. Liveness is an important feature that reflects physiological signs and differentiates artificial from real biometric traits. This paper aims to provide a simple path for the future development of more robust and accurate liveness detection approaches.

This paper discusses about introduction to the face biometric system, liveness detection in face recognition system and comparisons between the different discussed works of existing measures.

This paper presents an overview, comparison and discussion of proposed face liveness detection methods to provide a reference for the future development of more robust and accurate liveness detection approaches.

Health scientists and urban planners have long been interested in the influence that the built environment has on the physical activities in which we engage, the environmental hazards we face, the kinds of amenities we enjoy, and the resulting impacts on our health. However, it is widely recognized that the extent of this influence, and the specific cause-and-effect relationships that exist, are still relatively unclear. Recent reviews highlight the need for more individual-level data on daily activities (especially physical activity) over long periods of time linked spatially to real-world characteristics of the built environment in diverse settings, along with a wide range of personal mediating variables. While capturing objective data on the built environment has benefited from wide-scale availability of detailed land use and transport network databases, the same cannot be said of human activity. A more diverse history of data collection methods exists for such activity and continues to evolve owing to a variety of quickly emerging wearable sensor technologies. At present, no “gold standard” method has emerged for assessing physical activity type and intensity under the real-world conditions of the built environment; in fact, most methods have barely been tested outside of the laboratory, and those that have tend to experience significant drops in accuracy and reliability. This paper provides a review of these diverse methods and emerging technologies, including biochemical, self-report, direct observation, passive motion detection, and integrated approaches. Based on this review and current needs, an integrated three-tiered methodology is proposed, including: (1) passive location tracking (e.g., using global positioning systems); (2) passive motion/biometric tracking (e.g., using accelerometers); and (3) limited self-reporting (e.g., using prompted recall diaries). Key development issues are highlighted, including the need for proper validation and automated activity-detection algorithms. The paper ends with a look at some of the key lessons learned and new opportunities that have emerged at the crossroads of urban studies and health sciences.

We do have a vision for a world in which people can walk to shops, school, friends' homes, or transit stations; in which they can mingle with their neighbors and admire trees, plants, and waterways; in which the air and water are clean; and in which there are parks and play areas for children, gathering spots for teens and the elderly, and convenient work and recreation places for the rest of us. (Frumkin, Frank, & Jackson, 2004, p. xvii)

Vehicles estimation can be used in evaluating traffic conditions and facilitating traffic control, which is an important task in intelligent transportation system. The paper aims to propose a vehicle-counting method based on the analysis of surveillance videos.

The paper proposes a novel two-step method using low-rank representation (LRR) detection and locality-constrained linear coding (LLC) classification to count the number of vehicles in traffic video sequences automatically. The proposed method is based on an offline training to understand an LLC-based classifier with extracted features for vehicle and pedestrian classification, followed by an online counting algorithm to count the number of vehicles detected from the image sequence.

The proposed method allows delivery estimation (counting the number of vehicles at each frame only) and total number estimation of vehicles shown in the scene. The paper compares the proposed method with other similar methods on three public data sets. The experimental results show that the proposed method is competitive and effective in terms of computational speed and evaluation accuracy.

The proposed method does not consider illumination. Hence, the results might be unsatisfactory under low-lighting condition. Therefore, researchers are encouraged to add a term that controls the illumination changes into the energy function of vehicle detection in future work.

The paper bridges the gap between LRR detection and vehicle counting by taking advantage of existing LLC classification algorithm to distinguish different moving objects.

The purpose of this paper is to present a procedure for finding unusual patterns in accredited tests using a rapid processing method for analyzing video records. The procedure uses the temporal differencing technique for object tracking and considers only frames not identified as statistically redundant.

An accreditation organization is responsible for accrediting facilities to undertake testing and calibration activities. Periodically, such organizations evaluate accredited testing facilities. These evaluations could use video records and photographs of the tests performed by the facility to judge their conformity to technical requirements. To validate the proposed procedure, a real-world data set with video records from accredited testing facilities in the field of vehicle safety in Brazil was used. The processing time of this proposed procedure was compared with the time needed to process the video records in a traditional fashion.

With an appropriate threshold value, the proposed procedure could successfully identify video records of fraudulent services. Processing time was faster than when a traditional method was employed.

Manually evaluating video records is time consuming and tedious. This paper proposes a procedure to rapidly find unusual patterns in videos of accredited tests with a minimum of manual effort.

The problem of automatic recognition of human activity is one of the most important and challenging areas of research in computer vision because of the wide range of possible applications, for example surveillance, advanced human‐computer interactions, monitoring. This paper presents statistical computer vision approaches to automatically recognize different human activities.

The human activity recognition process has three steps: firstly human blobs are segmented by motion analysis; then the human body posture is estimated and, finally a temporal model of the detected posture series is generated by discrete hidden Markov models to identify the activity.

The system was tested on image sequences acquired in a real archaeological site while some people simulated both legal and illegal actions. Four kinds of activity were automatically classified with a high percentage of correct detections.

The proposed approach provides efficient solutions to some of the most common problems in the human activity recognition research field: high detailed image requirement, sequence alignment and intensive user interaction in the training phase. The main constraint of this framework is that the posture estimation approach is not completely view independent.

Results of time performance tests were very encouraging for the use of the proposed method in real time surveillance applications.

The proposed framework can work using low cost cameras with large view focal lenses. It does not need any a priori knowledge of the scene and no intensive user interaction is required in the early training phase.