Extending an Application of CNN to Automate Leaf Area Measurement of Groundnut

Abstract

Groundnut, a vital crop known for its oil and protein content, ranks India as the second-largest producer globally. Enhancing the physical and gravimetric properties of groundnut is crucial for improving overall yield. One key biophysical metric for assessing crop growth is the Leaf Area Index (LAI), which relies on accurate leaf area measurement. Traditionally, the millimeter graph paper method-which is sometimes prone to errors-is used to measure the leaf area. This work investigates the use of machine learning techniques to automate the measurement of groundnut leaf area in order to aid in the estimation of the LAI. To improve the precision of estimating leaf area, a large dataset of groundnut leaf photos was collected from numerous field tests and pre-processed. In addition, a specialized hardware configuration was created to reduce human entry errors by automatically measuring the leaf area and sending data to a cloud server. To reduce the effect of lighting changes, the hardware system has homogeneous illumination This hardware configuration in conjunction with machine learning has proven to be successful in automating the measurement of leaf area and simplifying data administration. The suggested technique offers a dependable way to track crop growth and boost agricultural output because it can be adjusted to various crop varieties.