Maja Kocoń

Dr. Maja Kocoń

West Pomeranian University of Technology in Szczecin
orcid

Research Focus & Metrics

Academic Career Length

18Years (2009-Present)

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

Representative Publications

View All Publications
19 September 2025 in Applied Sciences
Verified
Journal Article
Open access
PDF file
MKMaja Kocoń
Maja Kocoń
SPSzymon Pawlukiewicz
Szymon Pawlukiewicz
This study addresses the task of simultaneous age estimation and gender classification from facial images using convolutional neural networks (CNNs). The objective was to develop a unified model capable of handling both regression and classification tasks effectively. Four models with varying architectures, loss functions, and preprocessing strategies were implemented and evaluated. The best-performing model achieved over 90% accuracy in gender classification and a mean absolute error (MAE) below four years for age estimation. Performance analysis showed variation across age groups, with reduced accuracy for elderly individuals due to dataset imbalance and improved predictions for younger and middle-aged adults. To assess generalization, the model was also tested on external images, maintaining strong performance, particularly in gender classification. Challenges such as overfitting and face misdetection were addressed through preprocessing adjustments and model tuning. Beyond empirical results, this work consolidates a unified, reproducible protocol for joint age estimation and gender classification on a widely used face database. We standardize preprocessing, implement a consistent image-level split with a published seed, and define task-appropriate metrics. All training details are documented to provide a baseline, enhanced by a qualitative error analysis, enabling consistent reporting and facilitating future extensions. This approach demonstrates the effectiveness of CNNs for age and gender prediction and highlights their potential integration into recommendation systems that personalize content based on demographic attributes.

29 July 2025 in Applied Sciences
Verified
Journal Article
Open access
PDF file
Barbara SzymanikBarbara Szymanik
Barbara Szymanik
MKMaja Kocoń
Maja Kocoń
Sam Ang KeoSam Ang Keo
Sam Ang Keo
FBFranck Brachelet
Franck Brachelet
DDDidier Defer
Didier Defer
Non-destructive evaluation of reinforced concrete structures is essential for effective maintenance and safety assessments. This study explores the combined use of active microwave thermography and deep learning to detect and localize steel reinforcement within concrete elements. Numerical simulations were developed to model the thermal response of reinforced concrete subjected to microwave excitation, generating synthetic thermal images representing the surface temperature patterns of reinforced concrete, influenced by subsurface steel reinforcement. These images served as training data for a deep neural network designed to identify and localize rebar positions based on thermal patterns. The model was trained exclusively on simulation data and subsequently validated using experimental measurements obtained from large-format concrete slabs incorporating a structured layout of embedded steel reinforcement bars. Surface temperature distributions obtained through infrared imaging were compared with model predictions to evaluate detection accuracy. The results demonstrate that the proposed method can successfully identify the presence and approximate location of internal reinforcement without damaging the concrete surface. This approach introduces a new pathway for contactless, automated inspection using a combination of physical modeling and data-driven analysis. While the current work focuses on rebar detection and localization, the methodology lays the foundation for broader applications in non-destructive testing of concrete infrastructure.

1 September 2014
Verified
Conference Paper
Closed access
MKMaja Kocoń
Maja Kocoń
Nowadays, in Human-Machine Interaction systems, the communication between user and the machine (robot, computer, etc.) may be performed using a graphical representation of human often called avatar. Such graphical form needs a realistic human motion synthesis mechanisms which are becoming more and more popular in the virtual environments. Therefore the paper is focussed on the problem of facial motion modeling for interactive virtual humans using events-based interaction rules. First, the most frequent emotional states and two types of personalities have been selected. For both personalities the transitions of the emotional states have been performed based on the motion probabilities. Then, for animation process a set of the facial expression subtypes has been introduced. In such case it is possible to get a quite natural reactions of the avatar for various emotional states. Also, the transition probabilities have been defined for sub-states. Having obtained transition probabilities, a temporal duration of emotional state have been calculated. Finally, the architecture of the system for interaction between user and avatar based on facial animation modeling has been proposed.

1 January 2009 in IFAC Proceedings Volumes
Verified
Journal Article
Closed access
PDF file
MKMaja Kocoń
Maja Kocoń
ZEZbigniew Emirsajłow
Zbigniew Emirsajłow

1 July 2018
Verified
Conference Paper
Closed access
MKMaja Kocoń
Maja Kocoń
This paper addresses the problem of simulating human head motions for avatars. The presented approach includes the data extracted in the video sequences analysis process, stochastic simulations and the three-dimensional geometrical head model. The main goal of the research concerns on expressing emotion through head movement. Therefore we have described the head movements when a person expresses facial expressions. The aspect of head movement is essential in avatar/robot motion, because it provides additional information about the emotional state or it will fill a gap in idle animation where the virtual person or robot is waiting for action. Therefore the technique can increase the level of realism while performing human-machine interactions.

23 September 2021
Verified
Conference Paper
Closed access
MKMaja Kocoń
Maja Kocoń
This paper focuses on rigid head motion synthesis for a virtual person in human–machine interaction systems. Head gestures transmit additional, nonverbal information in human communication and can be used to increase the emotional context of the facial animation authentically simulating human behaviour. The presented method includes the head motion of the expressive face. The proposed motion model was described using Lie algebra for four rigid objects, which reflect the character of the neck and head skeleton movement and jaw bone. The obtained dependencies were used to generate three-dimensional motion of the head model, which was made considering the anatomical structure of the human head. The description presented in this article has the main advantage that it can be extended with additionally rigid elements. It means that we can get a description of the movement of the entire human skeleton, which reflects human anatomy. The proposed motion description eliminates the requirement to use motion capture whenever we need avatar movement in a particular situation, and we use it only for measurement data acquisition.