A5087754464- Advanced Vision and Imaging
- Robotics and Sensor-Based Localization
- Medical Image Segmentation Techniques
- Advanced Image and Video Retrieval Techniques
- Neural Networks and Applications
- Image and Object Detection Techniques
- Face and Expression Recognition
- 3D Shape Modeling and Analysis
- Computer Graphics and Visualization Techniques
- Image Retrieval and Classification Techniques
- Human Pose and Action Recognition
- Optical measurement and interference techniques
- Algorithms and Data Compression
- Advanced Image Processing Techniques
- Video Surveillance and Tracking Methods
- Handwritten Text Recognition Techniques
- Image Processing and 3D Reconstruction
- Anomaly Detection Techniques and Applications
- Image and Signal Denoising Methods
- Robot Manipulation and Learning
- Image Processing Techniques and Applications
- Robotics and Automated Systems
- Wildlife-Road Interactions and Conservation
- Advanced Numerical Analysis Techniques
- Face recognition and analysis
Czech Technical University in Prague
2015-2024
Institute of Informatics of the Slovak Academy of Sciences
2014-2023
Nature Conservation Agency of the Czech Republic
2007-2022
PATH To Reading
2012
University of Ulster
2008
Centre de Recherche en Acquisition et Traitement de l'Image pour la Santé
2006-2008
Université Claude Bernard Lyon 1
2006-2008
TU Dortmund University
2008
First Technical University
2007
Halmstad University
2004
This paper presents a method for recognizing human actions based on pose primitives. In learning mode, the parameters representing poses and activities are estimated from videos. run can be used both videos or still images. For primitives, we extend Histogram of Oriented Gradient (HOG) descriptor to better cope with articulated cluttered background. Action classes represented by histograms sequences, incorporate local temporal context means n-gram expressions. recognition is simple histogram...
In this paper we describe a detector of facial landmarks based on the Deformable Part Models. We treat task landmark detection as an instance structured output classification problem. propose to learn parameters from data by Structured Output Support Vector Machines algorithm. contrast previous works, objective function learning algorithm is directly related performance resulting which controlled user-defined loss function. The real-time standard PC, simple implement and it can be easily...
This work presents a complete pipeline for folding pile of clothes using dual-armed robot. is challenging task both from the viewpoint machine vision and robotic manipulation. The presented comprised following parts: isolating picking up single garment crumpled garments, recognizing its category, unfolding series manipulations performed in air, placing roughly flat on table, spreading it, and, finally, it several steps. segmented into separate garments color texture information, ideal...
We propose a transformation from the multi-class support vector machine (SVM) classification problem to single-class SVM which is more convenient for optimization. The proposed based on simplifying original and employing Kesler construction can be carried out by use of properly defined kernel only. experiments conducted indicate that method comparable with one-against-all decomposition solved state-of-the-art sequential minimal optimizer algorithm.
The work addresses the problem of clothing perception and manipulation by a two armed industrial robot aiming at real-time automated folding piece garment spread out on flat surface. A complete solution combining vision sensing, segmentation understanding, planning its real execution is proposed. new polygonal model introduced. Fitting into segmented contour used to detect landmark points. It shown how folded variants unfolded can be derived automatically. Universality usefulness...
AbstractThis paper describes our experience in designing, developing and deploying systems for supporting human–robot teams during disaster response. It is based on R&D performed the EU-funded project NIFTi. NIFTi aimed at building intelligent, collaborative robots that could work together with humans exploring a site, to make situational assessment. To achieve this aim, addressed key scientific design aspects up situation awareness team, using user-centric methodology involving end users...
We provide key facts about the TRADR project deployment of ground and aerial robots in Amatrice, Italy, after major earthquake August 2016. The were used to collect data for 3D textured models interior exterior two badly damaged churches high national heritage value.
Presents a new approach to rendering arbitrary views of real-world 3D objects complex shapes. We propose represent an object by sparse set corresponding 2D views, and construct any other view as combination these reference views. show that this can be linear, assuming proximity the we suggest how visibility constructed points determined. Our makes it possible avoid difficult reconstruction, only is required. Moreover, almost no calibration needed. present preliminary results on real objects,...
The development of sustainable transport is a key challenge in societies where there an accelerated need for socio-economic development. This the case seven countries from central and south-eastern Europe that share Carpathian Mountains. developing requires transdisciplinary, or at least cross-sectoral cooperation, between nature conservation sectors. Such cooperation not culture countries, which together host some most remarkable biodiversity values Europe, including largest populations...
In surgical practice, small metallic instruments are frequently used to perform various tasks inside the human body. We address problem of their accurate localization in tissue. Recent experiments using medical ultrasound have shown that this modality is suitable for real-time visualization anatomical structures as well position instruments. propose an image- processing algorithm permits automatic estimation a line-segment-shaped object. This method was applied thin electrode biological show...
We propose a novel MRF-based model for deformable image matching. Given two images, the task is to estimate mapping from one other maximizing quality of match. consider mappings defined by discrete deformation field constrained preserve 2D continuity. pose as finding MAP configurations pairwise MRF. more compact MRF representation problem which leads weaker, though computationally tractable, linear programming relaxation - approximation technique we choose apply. The number dual LP variables...