Creating a Comprehensive ML Pipeline for Weedbot

Worked on an impact-driven startup Weedbot, where we as a team of 40 people from different countries developed a laser weeding machinery for farmers that is able to localize plants and distinguish them from weeds and then remove the weeds with a laser beam.

Goal: To develop a high-speed plant image recognition neural network with a speed of 12ms per image or faster and recognition precision of 100-110% of crop polygon (which means up to 10% false positives are allowed). Technology: Computer Vision, Machine learning