Q & A with Amogh Joshi
GEMINI Team Spotlight
Quick Summary
- Learn more about the GEMINI team in our series of team member spotlights
In our last newsletter, we briefly introduced our project and its members. Today, we would like to highlight two of our team members and learn more about their experience. We start with Amogh Joshi, a student researcher working with Dr. Mason Earles.
Please introduce yourself: who are you, where are you from, and what is your educational/professional background?
AJ: My name is Amogh Joshi. I live in Princeton, New Jersey, and am currently a senior in high school. I’ve been working in the Plant AI and Biophysics Lab at UC Davis for over a year and a half, and I’ve been part of GEMINI since September. My research background and interests primarily lie in computer vision and deep learning, and how they can be applied to agricultural technology and robotics.
What is your role within GEMINI?
AJ: I’m working on automating the development of biophysical crop models from imagery and other sensor data of real-world African crops. This revolves around integrating traditional deep learning models with the Helios simulation software. The future goal is that a farmer would be able to take a few pictures of a certain crop, which would then be translated into a limitless repository of synthetic imagery to train a wide array of new models on, eventually helping them increase crop yield. I’m also involved with the creation of a centralized data framework for GEMINI, such that all data and information acquired by the team can be retrieved through a common and intelligent access point.
How did you end up working with GEMINI and what interests you most about this project?
AJ: I was working with Professor Mason Earles on assessing methods for training agricultural deep learning models with limited quantities of data, and standardizing the data we had into a large-scale public framework called AgML. As I began working with more novel methods —— ranging from finetuning models trained on synthetic data to incorporating a broader range of crops into my pretrained models —— he introduced me to the work in GEMINI, which is in many ways a new application of the methods that I used in the research I did previously.
Is there anything particularly exciting you are working on now or in the near future that you would like to share?
AJ: I’m trying to develop new methods to extract geometric or structural information about plants from 2-dimensional images of them. If this works, I’ll be one step closer to successfully being able to translate a simple image of a plant into a 3-D model of it. What do you like to do outside of work? AJ:I’m an avid reader, and a major travel enthusiast. I’m often reading either a biography or a mystery novel; or planning my latest trip out to a new corner of the world. Aside from developing technology for plants, I also grow a vegetable garden of my own.