| Abstract [eng] |
This thesis investigates the application of artificial intelligence methods in the analysis of dermatological images, with a particular focus on the development and evaluation of an algorithm designed for the localization and assessment of skin reactions. The aim of the study is to evaluate the ability of deep learning models to identify features of skin reactions and to compare the effectiveness of different model architectures. To assess the performance of the algorithms, the Intersection over Union (IoU) and Dice score metrics were employed, enabling a quantitative evaluation of the overlap between predicted segments and ground truth annotations. The analysis of experimental results provided insights into the accuracy of different model architectures and their suitability for practical applications. The study also examines how different datasets influence the models' ability to detect pathological skin changes. The thesis consists of the following main parts: literature review, data preparation and processing, model training, and experimental analysis. The literature review discusses deep learning methods and their applications in dermatology. The experimental section presents the performed computations, model comparisons, and analysis of the obtained results. The thesis concludes with a summary of the findings and final conclusions. |
