20.500.12556/RUL-186472 Prepoznavanje divjadi z uporabo prenosa učenja in modela YOLO Wildlife Recognition Based on Transfer Learning and the YOLO Model V diplomskem delu smo obravnavali prepoznavanje prostoživeče divjadi na slikah lovskih kamer z uporabo prenosa učenja in modela YOLOv11s. Namen dela je bil izdelati model, ki žival na sliki zazna, določi njen položaj z očrtanim pravokotni- kom in jo razvrsti v ustrezen razred. Preverili smo tudi, kako pretvorba dnevnih barvnih slik v črno-belo obliko vpliva na uspešnost zaznavanja in razlikovanje med srno in srnjakom. Pripravili smo lastno podatkovno zbirko dnevnih RGB in nočnih infrardečih slik, zajetih z lovskimi kamerami na območju Semiča v Beli krajini. Slike smo označili na platformi Roboflow ter podatkovno zbirko z augmentacijo razširili na 3014 slik. Z uporabo predhodno naučenega modela YOLOv11s smo v okolju Ka- ggle naučili dva modela. Prvi je uporabljal kombinacijo dnevnih RGB in nočnih IR slik, pri drugem pa so bile vse slike pretvorjene v črno-belo obliko. Modela sta uporabljala enake razrede in enako delitev podatkov. Model RGB in IR je na validacijski množici dosegel natančnost 0,771, pri- klic 0,672, mAP@0.5 0,803 in mAP@0.5:0.95 0,644. Črno-beli model je dosegel natančnost in priklic 0,756, mAP@0.5 0,777 ter mAP@0.5:0.95 0,621. Črno-beli model je dosegel višji priklic in višjo največjo vrednost F1, model RGB in IR pa višji vrednosti mAP ter boljše rezultate pri razlikovanju med srno in srnja- kom. Rezultati kažejo, da odstranitev barvnih informacij ni izboljšala skupne uspešnosti. Za zanesljivejšo praktično uporabo bi bilo treba podatkovno zbirko dodatno razširiti, uravnotežiti in preveriti na slikah z drugih lokacij. This thesis addresses the recognition of free-ranging wildlife in trail-camera im- ages using transfer learning and the YOLOv11s model. The objective was to develop a model capable of detecting an animal, locating it with a bounding box, and assigning it to the appropriate class. The study also examined how con- verting daytime colour images to grayscale affects detection performance and the distinction between female and male roe deer. A custom dataset containing daytime RGB and nighttime infrared images was collected with trail cameras in the Semiˇc area. The images were annotated using the Roboflow platform, and data augmentation increased the dataset to 3014 images. Two models based on a pretrained YOLOv11s network were trained in the Kaggle environment. The first model used a combination of daytime RGB and nighttime IR images, while all images used by the second model were converted to grayscale. Both models used the same classes and data split. On the validation set, the RGB and IR model achieved a precision of 0.771, a recall of 0.672, an mAP@0.5 of 0.803, and an mAP@0.5:0.95 of 0.644. The grayscale model achieved a precision and recall of 0.756, an mAP@0.5 of 0.777, and an mAP@0.5:0.95 of 0.621. The grayscale model achieved a higher recall and a higher maximum F1 score, whereas the RGB and IR model achieved higher mAP values and better results when distinguishing between female and male roe deer. The results indicate that removing colour information did not improve over- all performance. For more reliable practical use, the dataset should be expanded, balanced, and evaluated on images from additional locations. prepoznavanje divjadi zaznavanje objektov prenos učenja YOLOv11 lovske kamere računalniški vid wildlife recognition object detection transfer learning YOLOv11 trail cameras computer vision true false false Slovenski jezik Angleški jezik Diplomsko delo/naloga 2026-09-02 10:05:02 2026-09-02 10:05:18 2026-09-08 11:14:16 0000-00-00 00:00:00 2026 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 63233 290350339 Sever_Matej_-_Prepoznavanje_divjadi_z_uporabo_prenosa_ucenja_in_modela_YOLO.pdf Sever_Matej_-_Prepoznavanje_divjadi_z_uporabo_prenosa_ucenja_in_modela_YOLO.pdf 1 8477B174B0F4864D4FC41696F4BD4192 f59dc12e06cded81ca404a139ea91065af7c005cf6c1d9a80efaf4ddc48e927e 9be28ea0-a6a4-11f1-9b0d-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=245735 Fakulteta za elektrotehniko 0 0 0