Year: 2024
Type: article
Abstract: The project focuses on storm impact assessment on banana plantations using deep learning and image processing techniques. It leverages drone-acquired images to perform semantic segmentation to identify damaged and undamaged regions within the plantation. A pre-trained DeepLabV3 model with a ResNet-50 backbone is fine-tuned for this purpose. The segmented images are analyzed to count standing and fallen trees, estimate yield loss, and assess overall plantation health. To enhance accuracy, the approach integrates machine learning algorithms such as Cross-Entropy Loss, Adam Optimizer, and Connected Component Analysis. The system offers a fast, automated, and scalable solution for precision agriculture, enabling timely decision-making and disaster recovery planning.
Authors: Devaki Wale, Prajakta Mali, Snehal Darade, Janahvi Ghadage, Nikita Misal, P. S. Doshi
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
Open Access status: diamond
OpenAlex ID: W4404854694