Agricultural Field Demarcation Measurement and crop detection using machine learning algorithms

Project Title

Agricultural Field Demarcation Measurement and crop detection using machine learning algorithms

Project Area of Specialization Artificial IntelligenceProject Summary

A machine learning model that will be able to identify, demark and measure crop fields in the provided satellite or drone image. Additionally, the model will be able to identify the type of crop(s) present in the field. The model will be completely functional and deployed by the end of the project duration

Project Objectives

To develop a Machine Learning algorithm that:

• Marks field boundaries on map imagery

• Measures the size of field

• Estimates the type of crop currently growing in the field

• Create a database where data relevant to field boundaries, size of field and type of crop is stored

Project Implementation Method

The algorithms will be developed using Python programming language (Libraries like tensorflow and opencv will be used). The data for model development and training will be provided by Farmdar and will consist of high-res satellite images. Cloud services will be used to maintain databases of the gathered data.

Benefits of the Project

The motivation for this project comes from the fact that we usually end up underestimating or/and overestimating our crop production. This inaccuracy in estimation results in Pakistan usually importing more crop than it should or exporting less than it should. In both cases, the government, the economy and the farmers face heavy losses. This could all be avoided, had each farmer known the correct size of his/her field. This would help the farmers and the government in getting a more accurate estimation of the yield from each field. This, in turn, will help the government in accurately determining the import and export of each crop.

Technical Details of Final Deliverable

A webapp with a fully functional algorithm that can detect and mark field boundaries, measure field area and detect the type of crop. A UI where the end-user would be able to upload picture of map and/or use existing map data and see the desired field demarked and with proper labelling of crop type, health and measurement.

Final Deliverable of the Project Software SystemCore Industry AgricultureOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Responsible Consumption and ProductionRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
GPU (RTX 3050)	Equipment	1	70000	70000
Laptop Connectors	Miscellaneous	1	10000	10000