Precision livestock farming increasingly relies on advanced object localization techniques to monitor livestock health and optimize resource management. In recent years, computer vision-based localization methods have been widely used for animal localization. However, certain challenges still make the task difficult, such as the scarcity of data for model fine-tuning and the inability to generalize models effectively. To address these challenges, we introduces COLO (COw LOcalization), a publicly available dataset comprising localization data for Jersey and Holstein cows under various lighting conditions and camera angles. We evaluate the performance and generalization capabilities of YOLOv8 and YOLOv9 model variants using this dataset. Our analysis assesses model robustness across different lighting and viewpoint configurations and explores the trade-off between model complexity, defined by the number of learnable parameters, and performance. Our findings indicate that camera viewpoint angle is the most critical factor for model training, surpassing the influence of lighting conditions. Higher model complexity does not necessarily guarantee better results; rather, performance is contingent on specific data and task requirements. For our dataset, medium complexity models generally outperformed both simpler and more complex models. Additionally, we evaluate the performance of fine-tuned models across various pre-trained weight initialization. The results demonstrate that as the amount of training samples increases, the advantage of using weight initialization diminishes. This suggests that for large datasets, it may not be necessary to invest extra effort in fine-tuning models with custom weight initialization. In summary, our study provides comprehensive insights for animal and dairy scientists to choose the optimal model for cow localization performance, considering factors such as lighting, camera angles, model parameters, dataset size, and different weight initialization criteria. These findings contribute to the field of precision livestock farming by enhancing the accuracy and efficiency of cow localization technology. The COLO dataset, introduced in this study, serves as a valuable resource for the research community, enabling further advancements in object detection models for precision livestock farming. / Master of Science / Cow localization is important for many reasons. Farmers want to monitor cows to understand their behavior, count cows in a scene, and track their activities such as eating and grazing. Popular technologies like GPS or other tracking devices need to be worn by cows in the form of collars, ear tags etc. This requires manually putting the device on each cow, which is labor-intensive and costly since each cow needs its own device. In contrast, computer vision-based methods need only one camera to effectively track and monitor cows. We can use deep learning models and a camera to detect cows in a scene. This method is cost-effective and does not require strict maintenance. However, this approach still has challenges. Deep learning models need a large amount of data to train, and there is a lack of annotated data in our community. Data collection and preparation for model training require human labor and technical skills. Additionally, to make the model robust, it needs to be adjusted effectively, a process called model generalization. Our work addresses these challenges with two main contributions. First, we introduce a new dataset called COLO (COw LOcalization). This dataset consists of over 1,000 annotated images of Holstein and Jersey cows. Anyone can use this data to train their models. Second, we demonstrate how to generalize models. This model generalization method is not only applicable for cow localization but can also be adapted for other purposes whenever deep learning models are used. In numbers, we found that the YOLOv8m model is the optimal model for cow localization using our dataset. Additionally, we discovered that camera angle is a crucial factor for model generalization. This means that where we place the camera on the farm is important for getting accurate predictions. We found that top angles (placing the camera above) provide better accuracy.

The Generalization of the Digital Terrain Model Based on the TIN Abstract This diploma thesis deals with the up to now way and the possibilities of the digital terrain model generalization based on the TIN (the triangulate irregular network). New suitable way of the generalization of the digital terrain model procured from laser scanning data is proposed on the base of the existing generalization methods designated for digital models. Laser scanning data is characterized by a high areal density so the basic requirement is computing speed, maintaining the terrain features, such as a ridge, valley, steep hill, saddle, depression … and so on. The proposed algorithm is compared with the results of suggested algorithms and results from the generalization by the geographic software, such as Atlas DMT and ArcGIS.

Blended emotion is a classification of emotional experiences that involve the combination of multiple emotions. Research on the expression of blended emotions allows researchers to understand how different emotions interact and coexist in an individual’s emotional experience. Using machine learning to analyze mixed emotions may indeed bring new insights to the study of blended emotions. This thesis aims to explore blended emotion expression by testing machine learning models (SVM, Decision Tree, and Naive Bayes) trained on the single motion dataset on the blended emotion datasets and vice versa, to analyze the relationship between blended emotions and their constituent emotions. Furthermore, this thesis explores whether there is a dominant emotion in blended emotions and conducts an ablation study to investigate the importance of various facial features within each emotion. The results of testing models’ generalization capabilities propose that blended emotion expressions are highly likely to result from the overlapping combinations of features from their constituent emotions or the combination of some features from one constituent emotion with some from another. Furthermore, based on the dataset used, this thesis also finds that happiness predominated in the blended emotion ’disgust & happiness’. Additionally, an ablation study is conducted to identify the features that have the most significant impact on the accuracy and F1 score of single/pure emotion and blended emotion recognition across various recognition models. / ”Blandade känslor” är en klassificering av känslomässiga upplevelser som innefattar en kombination av flera känslor. Forskning om uttryck av blandade känslor möjliggör för forskare att förstå hur olika känslor interagerar och samexisterar i en individs känslomässiga upplevelse. Användningen av maskininlärning för att analysera blandade känslor kan faktiskt ge nya insikter i studiet av blandade känslor. Denna avhandling syftar till att utforska uttryck av blandade känslor genom att testa maskininlärningsmodeller (SVM, beslutsträd och Naive Bayes) som är tränade på dataset med enskilda känslor på dataset med blandade känslor och vice versa, för att analysera sambandet mellan blandade känslor och deras beståndsdelar. Dessutom utforskar denna avhandling om det finns en dominerande känsla i blandade känslor och genomför en ablationsstudie för att undersöka betydelsen av olika ansiktsdrag inom varje känsla. Resultaten av testning av modellernas generaliseringsförmåga föreslår att uttryck av blandade känslor sannolikt härrör från överlappande kombinationer av drag från deras beståndsdelar eller en kombination av vissa drag från en beståndsdel med vissa från en annan. Vidare, baserat på det använda datasetet, finner denna avhandling också att glädje dominerar i den blandade känslan ’avsky och glädje’. Dessutom genomförs en ablationsstudie för att identifiera de drag som har störst påverkan på noggrannheten och F1-poängen för igenkänning av enskilda/rena känslor och blandade känslor över olika igenkänningsmodeller.