Comparative study of Machine learning Approaches on Cropland Mapping

Due to heterogeneous and scattered land holdings, complicated crop sowing windows, and a lack of ground data, automated satellite-based crop mapping is still difficult in India. Effective approaches for mapping seasonal crop type must be developed while keeping in mind the large volume of satellite data. Currently the only approaches for crop mapping is in-person surveying or capturing data through aerial vehicles and mapping the areas of crops or RSGIS which again requires human intervention for cross-checking. Machine learning is identified as an approach for handling such a huge volume of data and its processing. Based on the current scenario of the domain, there exists a requirement of developing a system which accomplishes this task of cropland mapping. The overall objective of this work is to present the advantages of the machine learning approaches to the cropland mapping as compared to the traditional approaches and how it is beneficial for vast amounts of data which could make it easier to process using the machine learning approaches.