News

2014-12-16

Research visits in California

Prof. Kälviäinen visited UCB, UCSC, and Standford University

Professor Heikki Kälviäinen from MVPR visited University of California at Berkeley (UCB), University of California at Santa Cruz (UCSC), and Standford University December 3, 2014 - December 15, 2014. He visited several research groups related to digital image processing, machine vision, and pattern recognition, and gave presentations about MVPR's research actitivies. The goal of the visit was to continue currect joint research and to discuss about new possibilities for vision-based research collaboration.

The visit was hosted by Prof. Brian Barsky from UCB, Prof. Hartmut Sadrozinski from UCSC, and Prof. Fei Fei Li from Standford University.


2014-09-22

Visiting Lectures: Parallel Coordinates in Computer Vision and Object Detection with WaldBoost

Dr. Dubska and Dr. Juranek from Brno University of Technology introducing their research

Dr. Marketa Dubska and Dr. Roman Juranek from Brno University of Technology (BUT) from Czech Republic will give presentations about

Parallel Coordinates in Computer Vision and
Object Detection with WaldBoost

Wednesday September 24, 2014 at 14:30-16:00 in Room 2411 (seminar room at MVPR).

See the abstracts below.

Welcome!

Prof. Heikki Kälviäinen

Dr. Marketa Dubska: Parallel Coordinates in Computer Vision

The presentation focuses on usage of parallel coordinates in computer vision. The parallel coordinate system represents the space with axes which are mutually parallel. It can be used as a parameterization of lines and points for the Hough transform which requires only rasterization of lines, therefore it is very fast and accurate. The method is demonstrated on the matrix code extraction and the detection of vanishing points in Manhattan world images or roadside data.

Dr. Roman Juranek: Object Detection with WaldBoost

In my talk, I will briefly introduce WaldBoost algorithm for training object detectors (e.g. detection of faces in images). And I will summarize our research on acceleration of the detection process on various architectures (CUDA, FPGA). Finally, I will introduce our current research on multiview object detection with randomized forests, and its applications to traffic surveillance - detection of cars on highways.


2014-08-21

New Head of Degree Program

Prof. Lensu from MVPR selected as Head of Degree Program

Professor Lasse Lensu has been selected to serve as a head of Degree Program in Computational Engineering from the beginning of August.

The degree program is the youngest one in LUT and consists of three majors: Intelligent Computing, Technomathematics, and Technical Physics. Computational Engineering is a fascinating combination of mathematics, natural sciences, and computer science and engineering.

MVPR congratulates Prof. Lensu!


2014-08-12

Visiting Lectures: Machine Vision and Digital Image Processing Research at BUT

Prof. Zemcik and Dr. Seeman from Brno University of Technology introducing their research

Prof. Pavel Zemcik and Dr. Michal Seeman from Brno University of Technology (BUT) from Czech Republic will give presentations about

Video Processing, Action and Object Detection and
Human Vision and Computer Graphics

Wednesday August 20, 2014 at 10:00-12:00 in Room 2411 (seminar room at MVPR).

See the abstracts below.

Welcome!

Professor Heikki Kälviäinen

Prof. Pavel Zemcik
Image and Video Processing Research Group
Department of Computer Graphics and Multimedia (DCGM)
Faculty of Information Technology (FIT)
Brno University of Technology (BUT)
Czech Republic

Video Processing, Action and Object Detection

The talk addresses video processing algorithms with focus on action detection, object detection, and shortly also mentions other techniques, such as high dynamic range processing. Action detection, from the point of view of this talk is based on processing of video sequences using space-time features run though visual vocabulary, converted into BOW and classified using SVM. The main point of research is not improvement of the detection itself, although the solution is a state of the art one, but rather investigation of how the size of the video sequence being processed affects the detection rate and whether an optimal length exists and how it is dependent on the individual actions to be detected. Object detection in images is one of the key algorithms of computer vision. The talk will also address recent advances in object detection at FIT and its implementation in programmable hardware, specifically in FPGA. The implementation of WaldBoost, LBP and LRD feature extraction, memory layout, and prediction of neighborhood results will be given. HDR image acquisition from multiple exposure cameras and HDR image encoding present important problems in image acquisition. As for the HDR, current advances in HDR acquisition and encoding will be briefly presented as well as a possibility of exploitation of HDR in the above algorithms. Performance and results will be discussed as well and a future work outlined. Additionally, some brief information about current direction of research at FIT in object detection, HDR image acquisition, and other relevant direction.

Dr. Michal Seeman
Image and Video Processing Research Group
Department of Computer Graphics and Multimedia (DCGM)
Faculty of Information Technology (FIT)
Brno University of Technology (BUT)
Czech Republic

Human Vision and Computer Graphics

Human visual system is a complex and sophisticated network. It adapts to different object distance, luminance level, illumination hue, contrast or detail level. Our sight can deal with rapid light changes, immense contrasts or very dim light. On the other hand, some of the mechanisms are sources of optical illusions. Modern computer graphics is often inspired by human vision. Local contrast, global illumination or hue perception are perceived differently to the artificial image capture. Understanding of our vision helps to capture, process and reconstruct images more reliably.


2014-06-26

Research visits in Czech Republic

Prof. Kälviäinen visited CTU and BUT

Professor Heikki Kälviäinen from MVPR visited Czech Technical Universty (CTU) and Brno University of Technology (BUT) June 11, 2014 - June 20, 2014. He visited several research groups related to digital image processing, machine vision, and pattern recognition, and gave a presentation about MVPR's research actitivies. The goal of the visit was to continue currect joint research and to discuss about possibilities for vision-based research collaboration.

The visit was hosted by Dr. Radek Marik from Department of Telecommunications from CTU and by Associate Professor Adam Herout from Department of Computer Graphics and Multimedia from BUT. At BUT Professor Kälviäinen also served as a doctoral dissertation examiner for Marketa Dubska's defense.


2014-06-10

Erasmus teacher exchange in UK

Prof. Kälviäinen visited University of Surrey

Professor Heikki Kälviäinen from MVPR visited University of Surrey in UK May 25, 2014 - May 30, 2014. He gave Erasmus teachers exchange lectures and visited several research groups related to digital image processing, machine vision, and pattern recognition. The visit advanced joint research and collaboration in international education.

The visit was hosted by Prof. Josef Kittler from Centre for Vision, Speech and Signal Processing (CVSSP) from Department of Department of Electronic Engineering.


2014-06-06

TrafficVision project for monitoring condition of traffic signs

MVPR improving road maintenance using machine vision with Finnish Transport Agency (LIVI)

MVPR is expanding its application areas to road maintenance. The goal of the TrafficVision project is to support road maintenance, especially in the winter using digital image processing and analysis. The developed methods are used for Finnish Transport Agency (Liikennevirasto, LIVI) to improve their maintenance processes. This kind of machine vision solutions consist of the following tasks: imaging by a camera attached to a road maintenance vehicle, detection of a traffic sign, recognition of the detected traffic sign, and evaluation of the condition of the traffic sign. By automating current prartices, significant long-term financial savings are expected.

...

The project consists of the following personnell: Prof. Heikki Kälviäinen (principal investigator), Prof. Lasse Lensu, and research assistant Petri Hienonen.


2014-06-05

Jukka Lankinen defended his thesis successfully

Examination of Dr. Jukka Lankinen's doctoral dissertation

Jukka Lankinen succesfully defended his thesis

Local Features in Image and Video Processing - Object Class Matching and Video Shot Detection

Thursday May 22, 2014 at 12 in Auditorium 1382.

Congratulations to Jukka!

Reader, Dr. Krystian Mikolajczyk from University of Surrey from UK and Docent, Dr. Esa Rahtu from University of Oulu from Finland served as the opponents. Jukka Lankinen's supervisor Professor Joni Kämäräinen from Machine Vision and Pattern Recognition Laboratory (MVPR) was the custos.

Further information: doctoral dissertation and media release


2014-04-02

Visiting Lectures: Towards Realization of Mobile Devices for Semi-automatic Diagnosis of Melanoma from Skin

Dr. GholamHosseini from Auckland University of Technology introducing his research

Dr. Hamid GholamHosseini from Auckland University of Technology (AUT) from New Zealand will give a presentation

Towards Realization of Mobile Devices for Semi-automatic Diagnosis of Melanoma from Skin

Thursday April 10, 2014 at 15:00-16:00 in Room 2411.

Abstract:
Computer aided diagnosis of medical images can help physicians in better detecting and early diagnosis of many symptoms and therefore reducing the mortality rate. Realization of an efficient mobile device for semi-automatic diagnosis of melanoma would greatly enhance the applicability of medical image classification scheme and make it useful in clinical contexts. In this talk, digital image processing techniques as applied to skin lesion images will be discussed and efficient algorithms for filtering, segmentation, feature extraction and classification will be presented for discriminating melanoma from benign nevus images. Moreover, a platform for implementing the developed algorithms for realization of a hand held device for early diagnosis of melanoma will be discussed.

See the biography of the speaker below.

Welcome!

Professor Heikki Kälviäinen

Presenter: Dr. Hamid GholamHosseini received his PhD in Biomedical Engineering from Flinders University, Australia in 2002 and MSc in Electrical and Electronic Engineering from Tehran University, Iran. He has worked as a senior research lecturer and program leader of Biotronic Engineering at School of Engineering, Auckland University of Technology, New Zealand since 2000 and as a visiting professor at different Universities in USA, Germany, China and Sweden. He has published more than 88 book chapters, journal papers and conference publications. His research interests include biomedical signal and image processing, embedded and reconfigurable systems, teleolfaction and telehealthcare monitoring systems.


2014-04-02

Summer trainees selected

MVPR selected the summer trainees for 2014

MVPR selected Mikko Haavisto, Teemu Huovinen, and Kimmo Kerminen as summer trainees. Congratulations!

Head of MVPR, Prof. Heikki Kälviäinen is very happy see that also this year the selected trainess represent high-quality students of the degree program.