Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article

Degree

Master of Engineering Science

Program

Electrical and Computer Engineering

Supervisor

Pearce, Joshua M

Abstract

Agricultural productivity in strawberry cultivation was enhanced through the application of machine learning in this study. Traditional methods for detecting diseases and assessing ripeness in strawberries were identified as labor-intensive and error-prone, which limited farming efficiency and reduced crop yields. To address these challenges, it was hypothesized that advanced machine learning models incorporating attention mechanisms could significantly improve these tasks. The objective was to evaluate the effectiveness of various models by optimizing them for specific agricultural applications. Two datasets of strawberry images were augmented, and three pretrained models—Vision Transformer (ViT), MobileNetV2, and ResNet18—were fine-tuned. Data quality was improved through background removal and noise reduction, and weighted training was employed to manage imbalanced class distributions. The robustness of the models was further tested using synthetic data generated via Blender to simulate data scarcity. The results indicated that the ViT model achieved the highest accuracy and precision in identifying diseases and assessing ripeness. The models' effectiveness was further enhanced by the integration of attention mechanisms, and the potential for real-world agricultural applications was validated through the use of synthetic data. This research demonstrated that crop monitoring could be significantly improved with advanced machine learning models, particularly ViT, offering a promising tool for more sustainable and efficient strawberry cultivation. Future studies were recommended to expand these methods to other crops and integrate them into broader agricultural practices to enhance productivity and sustainability.

Summary for Lay Audience

The application of machine learning and computer vision in agriculture, specifically in strawberry cultivation, has opened new avenues for precision farming, crop monitoring, and disease detection. These technologies provide critical, real-time data that enable farmers to make informed decisions, optimizing resource use and improving crop management. Traditional methods, which rely on manual inspection for disease detection, are labor-intensive and subject to human error. By contrast, computer vision offers a more efficient solution by automating the detection of diseases, pests, and other issues, leading to timely interventions and improved crop quality.

In addressing the challenges specific to strawberry cultivation, such as the lack of large, annotated datasets and the variability in environmental conditions, this research utilized advanced imaging techniques. The study involved collecting two separate sets of strawberry images, which were then enhanced through resizing and augmentation to train three pretrained models: Vision Transformer (ViT), MobileNetV2, and ResNet18. To counteract the issue of imbalanced class distribution, a weighted training approach was adopted, which equitably distributed the impact across all classes during the training process.

Moreover, the study explored the integration of models like Swin Transformer to tackle more complex segmentation tasks, overcoming the limitations of standard ViT models which lack necessary segmentation heads for pixel-level classification. Through the strategic use of synthetic data and machine learning algorithms, the study aimed to provide robust solutions for strawberry disease identification and ripeness classification, enhancing the capabilities of farmers to monitor and manage their crops effectively.

Download

Share

COinS