E2E model from crop suggestion to harvesting time prediction in sugarcane plant utilizing machine learning and deep learning

Rubini, P. E. and Kavitha, P. (2024) E2E model from crop suggestion to harvesting time prediction in sugarcane plant utilizing machine learning and deep learning. Progress in Agricultural Engineering Sciences, 20 (1). pp. 233-270. ISSN 1786-335X

Preview

Text 446-article-p233.pdf - Published Version Download (7MB) | Preview

Abstract

The automated system for enhancing plant growth presents an innovative approach to optimize quality of sugarcane cultivation for four main sugarcane growing zones. It includes issues like recommendation of crops based on soil nutrients, diagnosis of disease in the leaf and stem images of sugarcane, weed detection and harvesting time prediction. The research work proposed in the article presents an innovative two-stage approach for object detection and classification in agricultural imagery. Initially, YOLOv8 (You Only Look Once) is employed to accurately detect objects within images, delineating them with precise boundary boxes. Subsequently, the focus of hybrid model integrating Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, known as Contextual Long Short-Term Memory (CLSTM), is employed. This dual-stage methodology harnesses the speed and accuracy of YOLOv8 for robust object localization, while the CLSTM model ensures nuanced classification, contributing to comprehensive and accurate approach for object detection and crop-weed differentiation in agricultural scenarios. The proposed approach is compared with the four DL algorithms for identifying weeds in sugarcane crops and subsequently assessed their accuracy and F1 score performance. At a learning rate of 0.002, the findings of CLSTM showcase superior precision at 98.5%, recall at 97.8%, F1 score at 98.1%, and an overall accuracy of 97.7%. The subsequent task is harvesting time prediction, which entails identifying the best time to harvest sugarcane based on the planting period, weather predictions, and sugarcane brix value. The implementation of this automated system not only enhances the productivity of sugarcane cultivation but also serves as a model for sustainable and resource-efficient agriculture.

Actions (login required)

Edit Item