Author: NCC Admin

Posted on

IEEE IT2023: Deep learning-based vehicle speed estimation using the YOLO detector and 1D-CNN

Researchers from NCC Montenegro presented a paper at the 27th IEEE Conference on Information Technology IT2023. The paper is titled “Deep learning-based vehicle speed estimation using the YOLO detector and 1D-CNN” and authored by Andrej Cvijetić, Slobodan Djukanović and Andrija Peruničić

ABSTRACT : This paper addresses vehicle speed estimation using visual data obtained from a single video camera. The proposed method accurately predicts the speed of a vehicle, using the YOLO algorithm for vehicle detection and tracking, and a one-dimensional convolutional neural network (1D-CNN) for speed estimation. The YOLO algorithm outputs bounding boxes around detected objects in an image, which is, in our case, the vehicle whose speed is to be predicted. As input to our 1D-CNN speed estimation model, we introduce a novel feature based on the change of area of the bounding box around the vehicle. The feature, referred to as the changing bounding box area (CBBA), is obtained by calculating the area of the bounding box, frame-to-frame, as the vehicle approaches the camera. The shape of the CBBA curve remains closely the same for all vehicles, with differences conditioned by the value of the observed vehicle’s speed. The proposed method is trained and tested on the VS13 dataset. Experiments show that it is able to accurately predict the vehicle’s speed with an average error of 2.76 km/h, with the best performing vehicle having the average error of just 1.31 km/h. The proposed method exhibits the robustness as a key advantage, eliminating the need for prior knowledge of real-world dimensions such as the vehicle size, road width, camera distance and angle in relation to the road etc.

Link : https://ieeexplore.ieee.org/document/10078518

Posted on

IEEE IT2023: Vehicle Speed Estimation From Audio Signals Using 1D Convolutional Neural Networks

Researchers from NCC Montenegro presented a paper at the 27th IEEE Conference on Information Technology IT2023.

The paper is titled “Vehicle Speed Estimation From Audio Signals Using 1D Convolutional Neural Networks” and authored by Ivana Čavor and Slobodan Djukanović.

ABSTRACT : This paper presents an approach to acoustic vehicle speed estimation using audio data obtained from single-sensor measurements. One-dimensional convolutional neural network (1D CNN) is used to estimate the vehicle’s speed directly from raw audio signal. The proposed approach does not require manual feature extraction and can be trained directly on unprocessed time-domain signals. The VS13 dataset, which contains 400 audio-video recordings of 13 different vehicles, is used for training and testing of the proposed model. Two training procedures have been evaluated and tested, one based on determining optimal number of training epochs and the other based on recording model state with minimal validation loss. The experimental results show that the average estimation error on VS13 is 9.50 km/h and 8.88 km/h, respectively.

Link : https://ieeexplore.ieee.org/document/10078724

Posted on

Journal Technology and Health Care: Verification of temperature, wind and precipitation fields for the high-resolution WRF NMM model over the complex terrain of Montenegro

Researchers from NCC Montenegro presented a paper at the “Special Issue for magazine Technology and health care: official journal of the European Society for Engineering and Medicine”.  The paper titled “Verification of temperature, wind and precipitation fields for the high-resolution WRF NMM model over the complex terrain of Montenegro” is written by Zečević Aleksandar, Filipović Luka and Marčev Angel.

ABSTRACT :

BACKGROUND: The necessity of setting up high-resolution models is essential to timely forecast dangerous meteorological phenomena.

OBJECTIVE: This study presents a verification of the numerical Weather Research and Forecasting non-hydrostatic Mesoscale Model (WRF NMM) for weather prediction using the High-Performance Computing (HPC) cluster over the complex relief of Montenegro.

METHODS: Verification was performed comparing WRF NMM predicted values and measured values for temperature, wind and precipitation for six Montenegrin weather stations in a five-year period using statistical parameters. The difficult task of adjusting the model over the complex Montenegrin terrain is caused by a rapid altitude change in in the coastal area, numerous karst fields, basins, river valleys and canyons, large areas of artificial lakes on a relatively small terrain.

RESULTS: Based on the obtained verification results, the results of the model vary during time of day, the season of the year, the altitude of the station for which the model results were verified, as well as the surrounding relief for them. The results show the best performance in the central region and show deviations for some metrological measures in some periods of the year.

CONCLUSION: This study can give recommendations on how to adapt a numerical model to a real situation in order to produce better weather forecast for the public.

Link : https://content.iospress.com/articles/technology-and-health-care/thc229016

Posted on

Appreciative feedback on Parallel Computing Course

Parallel Programming training course was organised by UDG and HPC NCC Montenegro in cooperation with NCC Germany, from 8th November to 14th of December 2022. The training was dedicated both to companies interested in the parallel programming skills and to students eager to learn on theoretical basis and practical features of parallel computing, with 54 attendees registered in total. Program course covered: Concepts of parallel computers – purpose, architecture, division; Practical guidelines for the development of parallel programs based on the architecture of shared and distributed memory as well as on the hybrid model; Analysis of the performance of parallel programs including decomposition of serial program and transformation into parallel programs. Beside the theoretical part, the training also included practical examples, use cases and hands-on exercises that allowed participants to apply and test their parallel programming knowledge on supercomputing systems/HPC-working environment. Participants learned to identify parallelization problem, analyse parallel programs complexity and efficiency, and develop simple parallel program, with dedicated support of academic professors and HPC experts dr Luka Filipovic from NCC Montenegro and lecturers from The Leibniz Computing Center, NCC Germany.    

  

After the training, survey forms were sent out to regular participants, revealing interesting statistics on training activity, industry appeal and further expectations.

Regarding academic participants, this was an obligatory course for students of the MSc program Artificial Intelligence, but also 1/3 of students came from Engineering and IT faculty level programs. Regarding industry participants, the majority were coming from the ICT sector, with working experience up to 5 years (but also 38% with 15+ years).

Decision to attend Parallel Programming course was dominantly influenced by: 1) personal interest in developing parallel computing skills, 2) possibility to obtain practical experience and 3) engagement of international and experienced lecturers.

With regard to level of complexity, the majority of participants considered the program course demanding, but successfully managed. With regard to teaching program and course organisation, 65% and 71% of participants evaluated them with the highest grade, respectively. With regard to communication with lecturers, knowledgeable answers and useful consultations were highly appreciated. 86% of surveyed participants confirmed that the training course fully or mostly met their expectations. 

Being asked what they liked the most about the course, participants stated: hands-on approach, practical examples, expertise of lecturers and open source application used by NCC in Germany. And when it comes to possible improvements, focus was on providing more practical examples and tasks. Over 70% participants would be further interested in Artificial Intelligence, Machine Learning and Deep Learning training opportunities, and all industry representatives confirmed interest in potential cooperation with UDG on project activities.

The general aim of the training course was to increase the parallel programming skills in Montenegro, but also to promote EuroCC projects and supercomputing resources, and to encourage HPC-based project ideas and partnerships. 

Posted on

Participation of NCC Montenegro @ 3rd EUROCC/CASTIEL global conference

Sanja Nikolic, representative of HPC NCC Montenegro participated at 3rdEUROCC/CASTIEL global conference, organized with an aim of presenting of main results, key achievements and NCCs successful deliverables within EuroCC1 project phase.

EUROCC/CASTIEL representatives presented key highlights of EUROCC 1 project management with regard to Competence Map building; Training/Mentoring /Twinning activities; Industrial interaction support and Awareness Creation events and actions.

In the second part of the conference NCCs Finland, Luxembourg, Montenegro, Slovenia, Sweden and Turkey presented key achievements and EUROCC1 contributions in their respective working packages and selected project activities. NCC Montenegro presented multiple activities and overall results in the project segment related to HPC/HPDA/AI Trainings and Skills development.

Posted on

World-renowned professors and researchers gathered around “Deep Learning and HPC” training

University of Donja Gorica, EuroCC Montenegro (national competence center for supercomputers) and in cooperation with the National Center of Spain (NCC Spain) organize the course “Deep Learning and HPC”. Guest lecturers from renowned institutions such as the University of Cambridge, Universidad de Cantabria and Barcelona Supercomputing Centre, GraphCoreAI and Imperial College London, AI Clearing and Shanghai Jiao Tong University, China, DeepMind, Montreal Institute for Learning Algorithms, Max Planck Institute will be involved in the implementation of the training.

During this course students will learn how to implement deep learning models for real situations on their local machines, key mathematical concepts of deep learning, some key concepts about computer vision and natural language processing.

In the second part of the training, students will learn about high-performance computing in the development of deep learning applications, parallel computing using the Python language on multiple CPUs or GPUs, and the pyTorch library for developing deep learning models. The course starts with guest lectures at the end of November so you need to express your interest as soon as possible. The course is followed by students of the Master AI program at UDG, but also by students of other programs who express an interest in it. Lectures are open to academia, industry and the public sector.

Recommended prerequisites for successfully attending lectures are: good knowledge of algorithms and mathematics, knowledge of at least one programming language (Python is preferred), basic knowledge of high-performance computing (HPC), basic knowledge and interest in artificial intelligence, knowledge of the English language.

Media about us:

https://www.portalanalitika.me/clanak/svjetski-priznati-profesori-i-istrazivaci-okupljeni-oko-teme-duboko-ucenje-i-hpc-u-okviru-ncc-na-udg

https://www.cdm.me/drustvo/svjetski-priznati-profesori-i-istrazivaci-okupljeni-oko-teme-duboko-ucenje-i-hpc-u-okviru-ncc-na-udg/

Posted on

Meeting with NCC Luxembourg

HPC NCC Montenegro representatives Sanja Nikolic and Luka Filipovic organized initial meeting with colleagues from NCC Luxembourg (representatives of Luxinnovation – National Innovation Agency; LuxProvide – in charge of long-term operations of national supercomputer – MeluXina, and University of Luxembourg – providing HPC educational programs and training courses) and discussed cooperation possibilities and best practices sharing in the areas of industry engagement, MeluXina features/access opportunities and training programs.

NCC Luxembourg – NCC Montenegro meeting

With regard to industry cooperation, colleagues from NCC Luxemburg shared their experience and expertise related to business users engagement and MeluXina onboarding process, focusing on HPC needs and benefits, compute-intensive workloads, industry-centric use cases and business cases, supported by dedicated sales & engineering teams. Petascale supercomputer, MeluXina is enabling leading research (non-commercial access through Euro HPC Access Calls) and industrial applications (65% of dedicated resources), with very high security and data protection standards highly appreciated by their commercial users. University of Luxemburg is organizing HPC School and HPC-related Master and Doctoral programs, covering various training needs from beginners access to advanced lectures, on its own and MeluXina supercomputing infrastructure. They also successfully manage HPC community of around 1000 users/ 200 newcomers per year.

NCC Montenegro representatives provided short introduction on main activities and achievements. Based on productive meeting, industry potential discussion and elaborated skills development programs, NCC teams agree on further alignment on HPC+ Topics & Trainings of joint interest, in the context of EuroCC2 plans and 2023 roadmap.

Privacy Policy Proudly powered by WordPress