Human pose estimation (HPE) has attracted a significant amount of attention from the computer vision community in past decades. Moreover, HPE been applied to various domains, such as human-computer interaction, sports analysis, and human tracking via images videos. Recently, deep learning-based approaches have shown state-of-the-art performance HPE-based applications. Although achieved remarkable HPE, comprehensive review methods remains lacking literature. In this article, we provide an...

Recently, memory-based methods have achieved progress in semi-supervised video object segmentation. However, these still suffer from unstructured challenges, such as transformations, occlusions and disappearance-reappearance. To this end, we propose a Holistic Correction Network (HCNet) to adaptively acquire concise prototypes for holistic correction at semantic, spatial temporal aspects. Specifically, an Adaptive Prototype Update module is firstly designed construct multi-level core...

The online extrinsic parameters calibration among multiple sensors for data fusion has been a crucial task in autonomous driving. This paper proposes new online, robust pipeline of automatic between monocular RGB cameras and LiDARs driving applications. novelties this work includes threefold: (1) an full the is developed system; (2) initialization stage without initial value parameters; (3) combination depth estimation edge detection used to improve accuracy robustness. We implement proposed...

Abstract Semantic web (SW) technology has been widely applied to many domains such as medicine, health care, finance, geology. At present, researchers mainly rely on their experience and preferences develop evaluate the work of SW technology. Although general architecture (e.g., Tim Berners-Lee’s Web Layer Cake) was proposed years ago well-known, it still lacks a concrete guideline for standardizing development In this paper, we propose an index standardize ensuring that is designed well...

This paper presents binary sensor network for human activity recognition, specifically, whose performance is enhanced by using convolution neutral networks (CNNs). The goal of this research to develop a sensing system based on CNNs that can recognize various daily activities with minimum data acquisition. whole consists pyroelectric infrared (PIR) arrays, feature extractor, and classifier. All the sensory are converted into numbers preserve geometry motion information targets. In work, we...

The challenge of robotic reproduction – making new robots by recombining two existing ones has been recently cracked and physically evolving robot systems have come within reach. Here we address the next big hurdle: producing an adequate brain for a newborn robot. In particular, task targeted locomotion which is arguably fundamental skill in any practical implementation. We introduce controller architecture generic learning method to allow modular with arbitrary shape learn walk towards...

Multiclass classification is a fundamental and challenging task in machine learning. The existing techniques of multiclass can be categorized as (i) decomposition into binary (ii) extension from (iii) hierarchical classification. Decomposing set classifications that efficiently solved by using classifiers, called class binarization, which popular technique for Neuroevolution, general powerful evolving the structure weights neural networks, has been successfully applied to In this paper, we...

Small robots have numerous interesting applications in domains like industry, education, scientific research, and services. For most vision is important, however, the limitations of computing hardware make this a challenging task. In paper, we address problem real-time object recognition propose Fast Regions Interest Search (FROIS) algorithm to quickly find ROIs objects small with low-performance hardware. Subsequently, use two methods analyze ROIs. First, develop Convolutional Neural...

While Deep Neural Networks (DNNs) achieve state-of-the-art performance in many fields, e.g., object recognition, they rely on deep networks with millions or even billions of parameters. Accelerating DNNs by reducing the parameters is crucial for real-time recognition. This paper presents an evolutionary approach to evolve efficient that can be run Low-Performance Computing Hardware (LPCH) recognition fastest possible speed and accuracy more than 95%. achieves goal means two design choices....

We present a feasibility study on evolving controllers for group of wheeled robot predators that need to capture prey robot. Our solution method works by in simulation 100 generations, followed 10 generations real robots. The best are further evaluated their sensitivity the initial positions. results demonstrate practical this approach and give an indication time required develop good solutions predator-prey problem.

This paper proposes two methods for developing UAV based virtual reality (VR) systems: (1) using a head tracking system and (2) panoramic 360 degrees' field of view (FoV) camera array. The former can be used to achieve real-time experiences users. latter generate off-line VR contents. work focuses on three techniques: head-camera synchronization, array design (3) flight paths planning. developed have been validated through experiments that either provide users with satisfactory or convert...

This paper presents an encoded smart floor for multiple human localization, include binary sensor designing, space encoding and decoding scheme. system can localize a group of people, as well recognize associated scenarios, with high sensing efficiency low computational complexity. The novelty this work includes: (1) set code design deployment; (2) Bayesian inference based scheme in the context activity scenario recognition. proposed has been tested pressure sensors, experiment results have...

We generalize the well-studied problem of gait learning in modular robots two dimensions. Firstly, we address locomotion a given target direction that goes beyond typical undirected gait. Secondly, rather than studying one fixed robot morphology consider test suite different robots. This study is based on our interest evolutionary systems where both morphologies and controllers evolve. In such system, newborn have to learn control their own body random combination bodies parents. apply...

This paper presents a Bayesian approach to energy efflcient and data-efflcient target localization using binary sensor networks. The novelty of this work lies in that the methods channel coding (code design, encoding decoding) are used solve problem. First, networks constructed via Low density parity-check (LDPC) matrices. As result, observation space targets is partitioned into many units encoded set codes. Then, when move around, system measurements will be with those codes through OR...

This paper presents a virtual reality (VR) system for the real-time teleconference. We develop distributed depth sensors that can reconstruct 3D images of users, and create panorama image conference room in real time. As result, users at remote locations have teleconference environment by wearing VR headsets. The contributions this work include development two-level sensor calibration data fusion scheme; improvement quality through point-to-mesh conversion; sensing computing architecture....

This paper presents a novel automatic system that remote operations between the user arm and multi-rotor UAV robotic to achieve synchronizes actions. The contains three parts: two sets of IMU motion sensors, six-degrees-of-freedom attached UAV, ground station displays real-time videos sent back by UAV-based camera. targets perform in or hazardous environment, such as water sampling, leakage estimation, radiation measurement. A two-step control scheme is developed action synchronization for...

In this paper, we develop a quadrotor UAV based target tracking and recognition system, which includes an intelligent gimbal sub-system for accurate camera positioning fast image processing. A set of robust consensus-based algorithms are developed objects tracking, in addition to moving background processing techniques. neural network learning database is used improve performance. Moreover, Geographic Information System (GIS) provide geo-location, environmental, contextual information the...

Scientific publications present biological relationships but are structured for human reading, making it difficult to use this resource semantic integration and querying. Existing databases, on the other hand, well automated analysis, do not contain comprehensive knowledge. We devised an approach constructing knowledge graphs from these two types of resources applied investigate between pre-/probiotics microbiota-gut-brain axis diseases. To end, we created (i) a base, dubbed ppstatement,...

This paper addresses the problem of designing behavioural strategies for a group robots with specific task, capturing another robot. Our proposed approach is to employ "smart" prey pre-programmed strategy based on novel Gaussian model danger zones and use an evolutionary algorithm (EA) optimize predators' behavior. The EA applied in two stages: first simulation, then hardware real robots. best evolved robot controllers are further inspected compared by their robustness, i.e., performance...