Deep Learning-Based Vision System for Water Stress Classification of Lettuce in Pot Cultivation

The increase in global population has resulted in a corresponding rise in the need for food and water, highlighting the significance of optimizing the utilization of scarce agricultural resources. Water scarcity is a major issue that affects crop growth, yield, and quality. Traditional plant water stress detection methods are time-consuming, expensive, and impractical for large-scale applications. To address this, researchers are developing non-invasive vision-based techniques that detect water stress in plants accurately utilizing vision systems and deep learning. Applying vision systems and deep learning in agriculture can help improve traditional agricultural practices and advance technology in the agriculture sector. This work aimed to develop a MobileNetV2-SVM model to classify and detect the degree of water stress in lettuce. According to the findings, the model achieved an accuracy of 87.50%. Implementing the system to detect mild water stress in green ice lettuce resulted in a notable improvement in crop output and a decrease in water usage. This implies that implementing a vision-based deep learning model to detect and classify water stress in green ice lettuce has the potential to boost water productivity, achieving more yield with less water.