Dario Bojanjac

Prof. Dario Bojanjac

orcid

Research Focus & Metrics

Academic Career Length

16Years (2011-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
10 November 2022 in Remote Sensing
Verified
Journal Article
Open access
PDF file
MKMarin Kačan
Marin Kačan
FTFilip Turčinović
Filip Turčinović
DBDario Bojanjac
Dario Bojanjac
MBMarko Bosiljevac
Marko Bosiljevac
The availability of low-cost microwave components today enables the development of various high-frequency sensors and radars, including Ground-based Synthetic Aperture Radar (GBSAR) systems. Similar to optical images, radar images generated by applying a reconstruction algorithm on raw GBSAR data can also be used in object classification. The reconstruction algorithm provides an interpretable representation of the observed scene, but may also negatively influence the integrity of obtained raw data due to applied approximations. In order to quantify this effect, we compare the results of a conventional computer vision architecture, ResNet18, trained on reconstructed images versus one trained on raw data. In this process, we focus on the task of multi-label classification and describe the crucial architectural modifications that are necessary to process raw data successfully. The experiments are performed on a novel multi-object dataset RealSAR obtained using a newly developed 24 GHz (GBSAR) system where the radar images in the dataset are reconstructed using the Omega-k algorithm applied to raw data. Experimental results show that the model trained on raw data consistently outperforms the image-based model. We provide a thorough analysis of both approaches across hyperparameters related to model pretraining and the size of the training dataset. This, in conclusion, shows how processing raw data provides overall better classification accuracy, it is inherently faster since there is no need for image reconstruction and it is therefore useful tool in industrial GBSAR applications where processing speed is critical.

21 January 2015 in IEEE Transactions on Antennas and Propagation
Verified
Journal Article
Closed access
Zvonimir SipusZvonimir Sipus
Zvonimir Sipus
DBDario Bojanjac
Dario Bojanjac
TKTin Komljenovic
Tin Komljenovic
This communication presents an accurate and computationally efficient method for designing spherical lens antennas. The method is based on spherical-wave modal expansion of both the electromagnetic field scattered by a multilayer lens and the far-field radiation pattern of a feed antenna. The two expansions are connected by transforming the fields from the local to the global coordinate system, i.e., by determining the equivalent current sources of the spherical lens Green's function. An important benefit of this method lies in the simplicity of implementing various feeding antennas in the design process. To verify the method, the results of the analysis are compared with measured radiation patterns of Luneburg lens antennas. Finally, the design guidelines for optimization of the whole lens antenna system are also discussed in the communication.

1 September 2020
Verified
Conference Paper
Open access
Petar KolarPetar Kolar
Petar Kolar
FTFilip Turčinović
Filip Turčinović
DBDario Bojanjac
Dario Bojanjac
Due to the COVID-19 pandemic that bursted out this year, online education has become a main type of education and an entire education system was forced to make a switch from classrooms to World Wide Web. The impact of the students receiving education online for the entire semester on their final grade was analysed and the results are shown in the paper. These results are obtained by analysing the academic course Mathematical Analysis 2, held online at the Faculty of Electrical Engineering and Computing, University of Zagreb, Croatia, in the summer semester of 2020. Furthermore, the number of views of the lectures in relation to the time left until the final exam was checked in order to investigate the students’ studying habits. Unsurprisingly, the most noticeable effect here is the last minute cramming. Nevertheless, the overall results show the pros and cons of online education, as well as the huge potential that online education has, even when performed at the current state.

5 April 2024 in Sensors
Verified
Journal Article
Open access
PDF file
FTFilip Turčinović
Filip Turčinović
MKMarin Kačan
Marin Kačan
DBDario Bojanjac
Dario Bojanjac
MBMarko Bosiljevac
Marko Bosiljevac
Zvonimir SipusZvonimir Sipus
Zvonimir Sipus
In recent years, the development of intelligent sensor systems has experienced remarkable growth, particularly in the domain of microwave and millimeter wave sensing, thanks to the increased availability of affordable hardware components. With the development of smart Ground-Based Synthetic Aperture Radar (GBSAR) system called GBSAR-Pi, we previously explored object classification applications based on raw radar data. Building upon this foundation, in this study, we analyze the potential of utilizing polarization information to improve the performance of deep learning models based on raw GBSAR data. The data are obtained with a GBSAR operating at 24 GHz with both vertical (VV) and horizontal (HH) polarization, resulting in two matrices (VV and HH) per observed scene. We present several approaches demonstrating the integration of such data into classification models based on a modified ResNet18 architecture. We also introduce a novel Siamese architecture tailored to accommodate the dual input radar data. The results indicate that a simple concatenation method is the most promising approach and underscore the importance of considering antenna polarization and merging strategies in deep learning applications based on radar data.

27 August 2020 in TM - Technisches Messen
Verified
Journal Article
Closed access
PDF file
Petar KolarPetar Kolar
Petar Kolar
LBLovro Blažok
Lovro Blažok
DBDario Bojanjac
Dario Bojanjac
In the last decade or two, it seems that the trend of technological advance in NMR spectroscopy cannot follow the trend of desire to measure NMR samples with gradually lower response signals. Recently, an accurate noise model, based on the concept of noise figure, of the most sensitive part of the NMR spectroscopy system from the aspect of noise, which is its probe-to-spectrometer receiving chain, was introduced. The main purpose of this model is to optimize the used NMR spectroscopy system and, ultimately, enable measuring NMR samples with even lower response signals than the ones measured today. All the parameters of the NMR spectroscopy system, used in the introduced model, can be easily measured using vector network analyzers and noise figure meters, or can be found in the datasheets of the respective elements, except the spectrometer’s receiving chain noise figure. In this contribution, the process of spectrometer’s receiving chain noise figure measurement, performed using the Twice Power Method, is described. A block diagram representation of the spectrometer’s receiving chain is presented here, as well as its approximative model. The respective noise figure measurement results, which are also presented, explain the general tendency of using the mid-range of the spectrometer’s gain control level when performing the actual NMR measurements.

10 July 2012 in IEEE Transactions on Antennas and Propagation
Verified
Journal Article
Closed access
DBDario Bojanjac
Dario Bojanjac
ZSZvonimir Sipus
Zvonimir Sipus
Scattering from anisotropic circular-cylindrical objects is rigorously treated for oblique incidence. The system of coupled differential equations for axial components of electromagnetic field arising in this problem is semi-analytically solved: using modal expansion in the axial and circumferential directions and by applying the finite-difference scheme in the radial direction. Developed analysis method is used to characterize different invisibility cloaks and it is demonstrated that realizable cloaks, which were designed for normal incidence, work properly for incidence angles of up to ±10° away from the normal. For angles greater than ±10° away from the normal the cloak should be redesigned, and for that an efficient design approach is realized by merging the developed program with a global optimization algorithm.