FOLTYN, L. Vylepšení metod detekce a klasifikace poškození otisku prstu [online]. Brno: Vysoké učení technické v Brně. Fakulta informačních technologií. 2023.
Student followed up on the previous works and create solution for detection and classification of the specific damages (line damage and moisture). Without understanding and usage of the previous work this would not be possible. The evaluation is fairly thorough, the text part is slightly above average, the student's activity was high. Taking all that into account I propose an overall grade of B ( very good ).
| Kritérium | Známka | Body | Slovní hodnocení |
|---|---|---|---|
| Informace k zadání | I consider the assignment of this bachelor's thesis to be moderately difficult, as it is related to the previous work. It uses analysis of neural networks for detection and classification of various fingerprint artifacts and new simulations of the artifacts into synthetic fingerprints. It is closely related to the work in the STRaDe research group. The assignment is fulfilled and I am satisfied with the results. | ||
| Práce s literaturou | Student followed the supervisor's recommendations and found other sources of study on his own. All sources are relevant and their use is correct. | ||
| Aktivita během řešení, konzultace, komunikace | The student was very proactive mainly in the summer semester. He continuously and regularly consulted on the solution procedure. He was properly prepared for all the consultations. | ||
| Aktivita při dokončování | The work was completed slightly in advance, but I did not see its final version before submission. Nevertheless almost all my recommendations from previous versions were applied to the final version. | ||
| Publikační činnost, ocenění | I am not aware of any awards or the student's publishing activities. |
Overall, the results obtained in the thesis are satisfactory. However the detection of straight lines, did not yield the desired outcome. This acknowledgment highlights an area for further improvement and potential future research. It suggests that additional investigation, experimentation, or alternative approaches may be necessary to overcome the difficulties associated with straight line detection.
| Kritérium | Známka | Body | Slovní hodnocení |
|---|---|---|---|
| Náročnost zadání | The objective of the bachelor thesis is to conduct a comparative analysis of existing object detection models and propose improvements for the detection and classification of line damage and moisture in fingerprint images. The requirements are met to obtain optimal results. | ||
| Rozsah splnění požadavků zadání | The damage artifacts were carefully selected and compared within the scope of the research. Processing methods were developed and implemented to address the detection and analysis of these damage artifacts. The precision of metrics was evaluated and measured to assess their effectiveness in accurately identifying and characterizing the damage. | ||
| Rozsah technické zprávy | The technical report has been effectively organized and structured to fit within the available space. | ||
| Prezentační úroveň technické zprávy | 79 | The content has been carefully presented concisely. The report effectively communicates the key aspects of the work while optimizing the use of space to deliver. | |
| Formální úprava technické zprávy | 79 | The thesis demonstrates a commendable standard in terms of typographical and linguistic aspects. The text is well-structured with clear headings, subheadings, and appropriate use of formatting elements such as font styles and sizes. | |
| Práce s literaturou | 77 | All relevant materials were met. The citations within the thesis are accurate and followed the appropriate referencing style. The results of the thesis work have been achieved, aligning with the objectives and research questions outlined in the study. | |
| Realizační výstup | 78 | The implementation of the solution showcases an understanding of the subject matter and the ability to apply technical knowledge effectively. The quality of the solution is commendable. The proposed solution is well-designed, taking into account relevant factors and considerations. | |
| Využitelnost výsledků | The work presented in the thesis builds upon previous implementations and leverages advanced models such as Faster R-CNN ResNet50, Faster R-CNN ResNet101, and CenterNet ResNet101 to enhance the results. By utilizing these state-of-the-art models, the research aims to improve the accuracy and efficiency of the object detection task. This continuation and integration of previous work, along with the utilization of advanced models, showcases a progressive approach in addressing the research objectives and contributes to the advancement of object detection techniques in the field of diseased fingerprint detection. |
eVSKP id 148340