In the case of precision agriculture, the monitoring of crop health indicators is required to maximize production and efficiently use resources. This paper proposes a novel approach to measure the Crop Water Stress Index (CWSI) in chilli plants. The method uses a deep learning segmentation algorithm on sunlit leaves collected from Kerala, a state in the southern part of India using a Fluke camera. The method can accurately identify water stress by analyzing the thermal images of plants and extracting specific characteristics that are indicative of the situation, such as segmented sunlight leaves. This research promotes precision agriculture by identifying the stress condition of chilli plants using CWSI values and conducting an empirical analysis of the chilli dataset. The analysis demonstrates that CWSI values effectively reflect varying levels of water stress in chili plants: mild stress (CWSI 0.0–0.4), moderate stress (CWSI 0.4–0.7), and high stress (CWSI above 0.7).This finding allows farmers to make more informed irrigation scheduling decision, enhancing agricultural productivity. This method has proven to be efficient in places with water scarcity and thereby increasing the profit margin of the farmers.