This study proposes the inflated modified Kumaraswamy (iMK) distribution, a flexible probability model defined on the unit interval [0,1]. It captures asymmetric behaviors while accommodating inflation at zero, one, or both boundaries, as commonly observed in normalized difference vegetation index (NDVI) data. Based on the iMK distribution, we develop a new regression model (iMKreg) suitable for double-bounded responses. From this model, we derive a detection tool for invasive plant species, particularly applicable to NDVI imagery. Model performance was evaluated using synthetic NDVI data, with further assessment of predictive accuracy and detection efficacy conducted on real-world measured NDVI image. The application to detecting black-grass (Alopecurus myosuroides) in wheat crops in southern Sweden shows that the iMKreg model outperforms both standard Gaussian-based linear regression and existing inflated Kumaraswamy regression models. 

Precision agriculture utilizes Artificial Intelligence (AI) and remote sensing to address challenges in sustainable food production. A significant obstacle is effective weed management, particularly for species like black-grass (Alopecurus myosuroides), which visually resembles wheat. This thesis introduces AI-driven methods using Unmanned Aerial Vehicle (UAV)-based multispectral image analysis to improve weed detection, addressing key challenges such as limited datasets, sensor misalignment, and the need for robust algorithms.

The first part of the thesis addresses data scarcity using synthetic data augmentation. A study using a Mask Region-based Convolutional Neural Network (Mask R-CNN) model for black-grass detection shows that scaling foreground objects is a particularly effective technique for improving performance, providing a basis for adapting models to new field conditions with less data.

The second part focuses on multispectral data quality, where sensor misalignment can corrupt Vegetation Indices (VIs) like Normalized Difference Vegetation Index (NDVI). A method for sensor calibration and image registration is presented to correct for these errors. The study then evaluates various VIs, finding that models integrating indices like Triangular Greenness Index (TGI) and Excess Green Index (ExG) show improved detection performance compared to using only the basic spectral bands, particularly between crop rows.

Finally, the third part draws parallels to biometrics to explore the robustness of different network architectures. By evaluating triplet-based and Siamese networks for fingerprint recognition, it is demonstrated that clustering-based approaches using contrastive loss offer better performance with incomplete and variable data. This insight highlights broader principles for creating robust recognition systems.

Overall, this thesis contributes methods for addressing key obstacles in precision agriculture, spanning data augmentation, sensor calibration, and the optimization of deep learning models. The findings of this work aim to contribute to the ongoing development of more effective agricultural practices.

This study addresses the issue of black-grass, a herbicide-resistant weed that threatens wheat yields in Western Europe, through the use of high- resolution Unmanned Aerial Vehicles (UAVs) and synthetic data augmentation in precision agriculture. We mitigate challenges such as the need for large labeled datasets and environmental variability by employing synthetic data augmentations in training a Mask R-CNN model. Using a minimal dataset of 43 black-grass and 12 wheat field images, we achieved a 37% increase in Area Under the Curve (AUC) over the non-augmented baseline, with scaling as the most effective augmentation. The best model attained a recall of 53% at a precision of 64%, offering a promising approach for future precision agriculture applications. © NLDL 2024. All rights reserved.

This work investigates the impact of the neural networks architecture when performing fingerprint recognition. Three networks are studied; a Triplet network and two Siamese networks. They are evaluated on datasets with specified amounts of relative translation between fingerprints. The results show that the Siamese model based on contrastive loss performed best in all evaluated metrics. Moreover, the results indicate that the network with a categorical scheme performed inferior to the other models, especially in recognizing images with high confidence. The Equal Error Rate (EER) of the best model ranged between 4%−11% which was on average 6.5 percentage points lower than the categorical schemed model. When increasing the translation between images, the networks were predominantly affected once the translation reached a fourth of the image. Our work concludes that architectures designed to cluster data have an advantage when designing an authentication system based on neural networks.

This paper presents a crucial multispectral image registration and sensor calibration method for an agricultural application. The multispectral images are obtained using a special drone equipped with multiple cameras flying at low altitudes. However, the distance between lenses, the lens distortions and the low-altitude flights lead to a lack of alignment in the built-in normalized difference vegetation index (NDVI). This lack of alignment results in a very poor performance in further analysis, especially for image segmentation and target detection to distinguish crops from invasive plants. In this work, we point out the importance of reducing this misalignment. To do so, the near-infrared and red sensors are first calibrated to remove the lens distortions. Then, the corresponding keypoints are utilized to calculate the transformation matrix and to minimize the back-projection error. The registered near-infrared and red images are then used to compute NDVI. The experimental results show higher alignment and F1-score of 0.73 which is a significant improvement in the performance of a trained deep neural network using NDVI in the detection of invasive plants. This is particularly a challenging task as the invasive plants resemble the desired crops.