This paper presents an application of a pedestrian-detection system aimed at localizing potentially dangerous situations under specific urban scenarios. The approach used in this differs from those implemented traditional systems, which are designed to localize all pedestrians the area front vehicle. Conversely, searches for critical areas only. environment is reconstructed with standard laser scanner, whereas following check presence performed due fusion vision system. great advantages such...

Microsoft Kinect had a key role in the development of consumer depth sensors being device that brought acquisition to mass market. Despite success this sensor, with introduction second generation, has completely changed technology behind sensor from structured light Time-Of-Flight. This paper presents comparison data provided by first and generation order explain achievements have been obtained switch technology. After an accurate analysis accuracy two under different conditions, sample...

This work presents a system based on an ensemble of Convolutional Neural Networks (CNNs) and descriptors for bioimage classification that has been validated different datasets color images. The proposed represents very simple yet effective way boosting the performance trained CNNs by composing multiple into combining scores sum rule. Several types ensembles are considered, with CNN topologies along learning parameter sets. not only exhibits strong discriminative power but also generalizes...

In 1979 Haralick famously introduced a method for analyzing the texture of an image: set statistics extracted from co-occurrence matrix. this paper we investigate novel sets descriptors matrix; in addition, compare and combine different strategies extending these descriptors. The following approaches are compared: standard approach proposed by Haralick, two methods that consider matrix as three-dimensional shape, gray-level run-length features direct use projected onto lower dimensional...

Objective. Mobile Brain/Body Imaging (MoBI) frameworks allowed the research community to find evidence of cortical involvement at walking initiation and during locomotion. However, decoding gait patterns from brain signals remains an open challenge. The aim this work is propose validate a deep learning model decode phases Electroenchephalography (EEG). Approach. A Long-Short Term Memory (LSTM) neural network has been trained deal with time-dependent information within EEG have preprocessed...

In this paper we propose a novel methodology for people re-identification based on skeletal information. Features are evaluated the skeleton joints and highly distinctive compact feature-based signature is generated each user by concatenating descriptors of all visible joints. We compared number state-of-the-art 2D 3D feature to be used with our two newly acquired public datasets RGB-D sensors. Moreover, tested approach against best methods in literature widely video surveillance dataset....

This paper presents the TerraMax vision systems used during 2007 DARPA Urban Challenge. First, a description of different is provided, focusing on their hardware configuration, calibration method, and tasks. Then, each component described in detail, algorithms sensor fusion opportunities: obstacle detection, road marking vehicle detection. The conclusions summarize lesson learned from developing passive sensing suite its successful fielding

Bioimage classification is increasingly becoming more important in many biological studies including those that require accurate cell phenotype recognition, subcellular localization, and histopathological classification. In this paper, we present a new General Purpose (GenP) bioimage method can be applied to large range of problems. The GenP system propose an ensemble combines multiple texture features (both handcrafted learned descriptors) for superior generalizable discriminative power....

In recent years, the field of deep learning has achieved considerable success in pattern recognition, image segmentation, and many other classification fields. There are studies practical applications on images, video, or text classification. Activation functions play a crucial role discriminative capabilities neural networks design new "static" "dynamic" activation is an active area research. The main difference between that first class activations considers all neurons layers as identical,...

Recognizing human actions is crucial for an effective and safe collaboration between humans robots. For example, in a collaborative assembly task, workers can use gestures to communicate with the robot, robot recognized anticipate next steps process, leading improved safety productivity. In this work, we propose general framework action recognition based on 3D pose estimation ensemble techniques, which allows recognize both body hand gestures. The relies OpenPose 2D lifting methods estimate...

CNNs and other deep learners are now state-of-the-art in medical imaging research. However, the small sample size of many data sets dampens performance results overfitting. In some areas, it is simply too labor-intensive expensive to amass images numbering hundreds thousands. Building Deep CNN ensembles pre-trained one powerful method for overcoming this problem. Ensembles combine outputs multiple classifiers improve performance. This relies on introduction diversity, which can be introduced...

This paper presents an application of a pedestrian detection system aimed at localizing potentially dangerous situations in specific urban scenarios. The approach used this work differs from the one implemented traditional systems, which are designed to localize all pedestrians appearing area front vehicle. first locates critical areas environment, and then it searches for these only. environment is reconstructed with standard laser scanner system, while following check presence performed...

We perform an extensive study of the performance different classification approaches on twenty-five datasets (fourteen image and eleven UCI data mining datasets). The aim is to find General-Purpose (GP) heterogeneous ensembles (requiring little no parameter tuning) that competitively across multiple datasets. state-of-the-art classifiers examined in this include support vector machine, Gaussian process classifiers, random subspace adaboost, rotation boosting, deep learning classifiers....

Industry 4.0 aims to make collaborative robotics accessible and effective inside factories. Human–robot interaction is enhanced by means of advanced perception systems which allow a flexible reliable production. We are one the contenders challenge with intent improve cooperation in industry. Within this competition, we developed novel visual servoing system, based on machine learning technique, for automation winding copper wire during production electric motors. Image-based often limited...

This paper describes a modular tracking system designed to improve the performance of pedestrian detector. The consists two modules, labeler and predictor. former associates identifier each pedestrian, keeping memory past history; this is achieved by merging detector predictor outputs combined with data about vehicle motion. predictor, basically Kalman filter, estimates new position observing his previous movements. Its output helps match between pedestrians detected in frame those observed...

The paper describes a human-robot cooperative installation methodology of heavy and bulky components based on marker-based visual servoing, force control, cooperation. main advance in the cooperation is achieved by shared-control interaction during task, relieving human operator manipulated load giving to robot partially autonomous behaviour force-tracking direction. Experimental results are shown context H2020 CleanSky 2 EURECA project which side-wall panel installed 1:1 scale mock-up...

This paper focuses on the key role played by adoption of a framework in teaching robotics with computer science approach master Computer Engineering.The adopted is Robot Operating System (ROS), which becoming standard de facto inside community.The educational activities proposed this are based constructionist approach.The Mindstorms NXT robot kit to trigger learning challenge.The ROS exploited drive students programming methodology during laboratory and allow exercise major paradigms best...

Within the Industry 4.0 context, industrial robots need to show increasing autonomy. The manipulator has be able react uncertainties/changes in working environment, displaying a robust behavior. In this paper, control framework is proposed perform interaction tasks uncertain scenes. methodology relies on two components: i) 6D pose estimation algorithm aiming recognize large and featureless parts; ii) variable damping impedance controller (inner loop) enhanced by an adaptive saturation PI...

Features play a crucial role in computer vision. Initially designed to detect salient elements by means of handcrafted algorithms, features now are often learned using different layers convolutional neural networks (CNNs). This paper develops generic vision system based on extracted from trained CNNs. Multiple combined into single structure work image classification tasks. The proposed was derived testing several approaches for extracting the inner CNNs and them as inputs support vector...