Category: Employee (student, worker)

Article: Towards Resource-Efficient DNN Deployment for Traffic Object Recognition: From Edge to Fog

Article: Towards Resource-Efficient DNN Deployment for Traffic Object Recognition: From Edge to Fog

The paper focuses on the challenges associated with deploying deep neural networks (DNNs) for the recognition of traffic objects using the camera of Android smartphones. The main objective of this research is to achieve resource-awareness, enabling efficient utilization of computational resources while maintaining high recognition accuracy. To achieve this, a methodology is proposed that leverages the Edge-to-Fog paradigm to distribute the inference workload across multiple tiers of the distributed system architecture. The evaluation was conducted using a dataset comprising real-world traffic scenarios and diverse traffic objects. The main findings of this research highlight the feasibility of deploying DNNs for traffic object recognition on resource-constrained Android smartphones. The proposed Edge-to-Fog methodology demonstrated improvements in terms of both recognition accuracy and resource utilization, and viability of both edge-only and edge-fog based approaches. Moreover, the experimental results showcased the adaptability of the system to dynamic traffic scenarios, thus ensuring real-time recognition performance even in challenging environments.

The link to conference can be found here: https://2023.euro-par.org/

The audio version is available:

 

FAAI at European Games Esports Championships

FAAI at European Games Esports Championships

The European Games Esports Championships (#EGE23) was accompanied by three prestigious substantive events, including an organized cyclical conference devoted to gaming and related industries in Central and Eastern Europe. Professionals, scientists and administrators shared their knowledge and experience in one place. The UBB team representative was taking part in a panel considered Gaming in education. He stressed the important part of AI in this this area. Additionally, the leaflets and poster were presented.

Article: Analysis of the Current Situation in Serbia Related to the Education in the Field of Applied Artificial Intelligence

Article: Analysis of the Current Situation in Serbia Related to the Education in the Field of Applied Artificial Intelligence

The fastest growing and most exciting scientific field today is Artificial Intelligence (AI) with its real world applications. The current transformation of society and business needs for AI specialists with specific competencies and skills dictate the transformation of the Serbian educational system and its adjustment toward modern demands. This paper presents some results and analysis of conducted surveys in the scope of the Erasmus+ project “Future is in Applied Artificial Intelligence” related to the current state of AI education in Serbia, existing university AI courses, knowledge and attitude of students and teachers toward AI contents, needs of the employers and the preferred future directions of the transformation of education system toward a competency-based digital society through formulating adequate AI learning requirements.

The article can be found at conference site: https://www.hm.kg.ac.rs/index.php

The audio version is available:

“Two Wheels” Rally

“Two Wheels” Rally

The “Two Wheels” Rally is an annual university action. On June 2, 2023, the 4th edition of the event took place. It was organized by the RESET IT scientific circle actively involved in the FAAI project. This year, the rally took place in the picturesque surroundings of Bielsko-Biała and Wilkowice, where green hills, winding paths, and charming places were the background for the cycling adventure of Erasmus+ FAAI project participants.
It was not only a rally, but it was also a great opportunity to integrate and make new friends with FAAI participants, university employees, and students. In addition, at the finish line of the rally – in the “Wilczysko” scout center in Wilkowice, the organizers prepared, as usual, numerous accompanying attractions, such as a draw for the main prize for students in the form of a bicycle, an off-road game, paintball, pea soup, and sausage.

Article: Exploring Image Unified Space for Improving Information Technology for Person Identification

Article: Exploring Image Unified Space for Improving Information Technology for Person Identification

This paper describes the study of conditions and parameter settings of the algorithm, that underlies in the basis of information technology of person recognition and identification, for creating a unified space for both low-quality and high-quality images with the purpose of developing the requirements to the input images, that will allow correctly identify images with different characteristics. For experimental research the algorithm was used that underlies in information technology of person identification in video stream. It includes the anisotropic diffusion as an image preprocessing method, Gabor wavelet transform as an image processing method, histogram of oriented gradients (HOG) and local binary patterns in 1-dimensional space (1DLBP) as the methods of feature vector extraction from the images. In this work anisotropic diffusion was applied with the parameter that preserves sharper boundaries than previous formulations and improves the automatic stopping of the diffusion. In the study three person images databases were used: The Database of Faces, Facial Recognition Technology (FERET) database and Surveillance Cameras Face Database (SCface). The algorithm performance provided various identification accuracy rate results with the difference, which amount to 20% in average. It arose the issue of input images space mismatching that significantly affected the algorithm performance. Therefore, it has been decided to perform the experimental research to search the possibility to create unified space of the requirements to the input images that will allow correctly identify images from different databases. So, there were performed the experiments with the image compression variety, difference of image resolutions and areas of face regions covering the images.

The paper can be found at: https://ieeexplore.ieee.org/document/10188824

The audio version is available: