Kalash Shah

Guiding Professor

Prof. Milind Sohoni

Project Title

Computational Analysis of Farmplot Polygons

Field

Computer Science

Problem Statement Description

Google is building a climate engine which recognises Farmplot boundaries using Computer Vision. They have minor discrepancies wrt ground truth and the farm records owned by the government. Our task was to develop algorithms for increasing the accuracy of predictions and aligning them with the government records.

What was your motivation for choosing this field?

I wanted to try a hands-on project in my sophomore summers in the CS field and this seemed a good learning opportunity with Google Research being involved.

What factors did you consider while choosing professors?

To be honest, I chose the project as this was my first experience. I knew the professor as he had taken our minor class and had briefed us about the project in the class.

Duration of the project

It was for 6 months from Jun ‘22 to Nov ‘22.

Overall experience

My work majorly involves execution and coding the algorithms developed by the professor. In the later stage, I was also involved in ideation as we tried out some Machine Learning options for increasing the accuracy.

The time commitment was higher in the vacations to the tune of about 15 hrs/week. In the semester it was about 6-10 hrs/week and manageable with the intern season.

I would say the project was moderate and did not involve complex research but testing of practical ideas. I liked the project as it was a great learning curve and later realised the importance of stakeholders involved in the project.

Key learnings

1. Good Amount of Python & SQL
2. Algorithms & ideation
3. Reporting and documenting your ideas clearly and in a formal manner. Yes the Project was the key talking point in my internship interviews and so I would say it definitely helped.

What course material had you gone through prior to the project?

I had taken CS213M, though there were no pre requisites.

Review of the Professor

The professor is strict when it comes to time commitment. One negative point was that he would message/call at any time even on Sunday evenings and had a slightly weird time cycle. Though I must admit he is quite smart and has good knowledge of the field.

Future plans following this project

The project would go on for 2-3 years as it is a comprehensive MoU with GoI and Google Research. Though I ended my part in Nov as the work seemed monotonous towards the end.

Anything else you feel is worth mentioning that was not covered in the questions above?

I would suggest to take reviews of the prof before taking project under him/her to avoid any unnecessary hastle at the end.