KALUTSKI, M. Rozhraní pro anotaci panelů fotovoltaických elektráren [online]. Brno: Vysoké učení technické v Brně. Fakulta informačních technologií. 2025.
Pan Kalutski konzultace prakticky téměř nevyužíval, takže nejsem schopen posoudit, jakým způsobem pracoval. Na naší poslední konzultaci ale nějaké výsledky předvedl, takže z implementačního hlediska na práci pracoval a technicky i něco vyřešil a nějaký výsledek vytvořil.
| Kritérium | Známka | Body | Slovní hodnocení |
|---|---|---|---|
| Informace k zadání | Zadání práce navazuje na loňskou bakalářskou práci, ve které byl vytvořen funkční detektor vad panelů fotovoltaických elektráren z termografických snímků. Tato práce si klade za cíl onen detektor rozšířit o propracované uživatelské rozhraní. Spolupráce s panem Kalutskim byla složitá, neboť v průběhu řešení prakticky nekomunikoval a konzultace nevyužíval. Pan Kalutski tak pracoval vesměs na vlastní pěst. | ||
| Práce s literaturou | Řešitel si sám vyhledával relevantní literaturu. | ||
| Aktivita během řešení, konzultace, komunikace | Pan Kalutski prakticky nekonzultoval, nereagoval ani na opakované výzvy ke konzultacím. Viděl jsem ho a měl možnost s ním konzultovat postup v práci v průběhu roku dohromady asi jen 3x. | ||
| Aktivita při dokončování | Finální obsah praktické části práce autor dořešil sám. Měl jsem možnost už jen zkontrolovat a poskytnout zpětnou vazbu k obsahu a úpravě technické zprávy. | ||
| Publikační činnost, ocenění |
The created application provides some of the functionality that is probably needed for the intended purpose. Althouhg, it is not complete, it could be extended into a useful tool. The biggest problem is the work with information sources. In my oppinion, the severity of the issues I discovered disqualifies the work from being accepted as bachalor thesis in the current state.
| Kritérium | Známka | Body | Slovní hodnocení |
|---|---|---|---|
| Náročnost zadání | |||
| Rozsah splnění požadavků zadání | In general, it sort of fits the criteria. However, I'm missing crucial functionality: Data and report exports Panel identification with respect to the plant layout. | ||
| Rozsah technické zprávy | |||
| Prezentační úroveň technické zprávy | 58 | The text is quite confusing. I had hard time understanding what was actually implemented and why. Information is often repeated. Requirements and use cases are not well defined - vague and incomplete. At some points, the text appears to contain clearly "wrong" information which does not represent the application at al: Table 5.2 suggests that Nginx is used to serve src/pages/* and src/components/* which is not true and would not make sense. It further states that backend (Application) provides "role-based ACL". I beleave that the backend does not provide any such functionality. Page 37 states that: "... supports scalability via stateless FastAPI pods replicated behind load-balancers, shared storeage via NFS, and containerised deployments ...". The application clearly does not provide such functionality. The text several times mentions that the "image" (orthophotomosaic) can be very large and is/should be tiled for rendering. I have not found any evidence that it is so. The application and design contain inconsistencies. Some of them are: How exactly does annotation versioning work? Why part of the text contains "inspectors" and "managers" and the application contains roles "Detector" and "Inspector"? Page 31 mentions "exportable dataset". The application does not provide export. | |
| Formální úprava technické zprávy | 77 | The text is written in good English and does not contain any serious typographic issues. | |
| Práce s literaturou | 40 | The work with literature is appalling. The previous thesis, which is the source of the backend implementation, is not cited. It is only mentioned in readme.md file. The text lacks sources in many, many, many places. Many sources are used incorectly - they do not contain the information they are used to source. Some of the the specific problems: I don't bother mentioning the many missing sources in the text. Page 5, [5] is used as a source for information about the use of Mask R-CNN, but it does not contain any mention of it. Page 11, information about GUIs, React, Angular, and web-based interfaces is sourced from [6], but it does not contain any information on these topics. Page 12, statements about React-based interfaces are sourced from [7], but it does not contain any mention about React. Page 14, [9] may appear to be the source for information about instance and semantic segmentation and U-net. However, it is only relevant to segmentation in general. Page 15, [11] probably does not contain much information on Usability. Page 16, [12] is used as a reference for Intuitiveness of user interfaces. However, it is the "Learning OpenCV" book. Page 17, statements about effective work with orthophotomosaic are sourced from [10] - a general book on UI/UX. No source for fault categories on Page 5. Probably missing sources for tables 2.1, 3.1, 3.2. | |
| Realizační výstup | 60 | The frontend application is a simple image annotation tool. I'm missing many functionalities I beleave are crucial for the intended use. Altogether it feels as an incomplete product. For example: the users are expected to manually check and correct plants with thousands of solar panels and it is imposible to mark which parts of the plant have already been checked and corrected! The testing is not convincing, clear or credible. The application is not ready for deployement. Frontend is served by a dev server. Backend may not be able to serve multiple users. I'm not sure where the application can be configured. No users No report compilation and export No raw annotation export No option to match solar panels with plant layout Very basic data visualization | |
| Využitelnost výsledků | The aplication is probably not usefull in the current state. |
eVSKP id 162844