Deep Learning Techniques for Region-Wise Crop Plantation Analysis in Konkan

Abstract

In this paper investigate the application of deep learning and remote sensing data for optimizing crop plantation in the Konkan region of Maharashtra. The practice of cultivating the soil, producing crops, and keeping livestock is referred to as farming. Agriculture is critical to a country’s economic development. Nearly 58 percent of a country’s primary source of livelihood is farming. Farmers till date had adopted conventional farming techniques. These techniques were not precise thus reduced the productivity and consumed a lot of time. Utilizing advanced frameworks like deep convolutional neural networks (DCNNs) and remote sensing techniques, this study focuses on soil, climate, and vegetation indices to enhance the precision of crop classification and yield prediction. The key objectives include evaluating environmental conditions for paddy and cultivation, mapping phenology using multispectral imaging, and proposing sustainable practices to address climate change impacts. By integrating Google Earth and DLR Earth Sensing Imaging Spectrometer data, this study aims to provide actionable insights into improving agricultural productivity and resilience in the region. To record paddy rice fields, we submitted a series of 19 NDVI images demonstrating crop phonological development into k-means clustering and Random Forest (RF) machine learning classification algorithms. Result shows that proposed model has overcome various evaluation parameters on different scale as compared to previous approaches adopted by researchers.