The rapid urbanization leads to more cropland loss, which makes land sustainability and food security planning much more challenging. Remote sensing can be an alternative approach to cropland mapping, especially in urban areas and provides important information about land use patterns, displacement of crops and encroachment. We propose a new method to map cropland in urban and peri-urban areas accurately using high-resolution satellite image, machine learning algorithms, and multi-temporal analysis. As the world population is expected to double, there will be a need for 70% more food than what we currently produce by 2050; therefore, new technologies like Remote Sensing and Machine Learning play a critical role in addressing these needs which are fundamental goals of United Nations SDG-2 “zero hunger”. For the number of applications for agricultural monitoring and food security, it is necessary to have a field-level crop map. More than 60% of the land in India is under cultivation. Before using satellite data to estimate agricultural productivity, mapping cropland is an important step. This study demonstrates how to fit harmonic regression and retrieve the coefficients to extract features from time series. Following that, croplands would be classified by random forest, and “ground truth” labels would be obtained from the ESA World Covers 2021. Cropland is being mapped using satellite data on a global scale, but mapping it locally remains a challenge.