Optimizing Growth Rate Counting with Computer Vision

In this project, I developed and productionized a cutting-edge computer vision solution to optimize the counting and growth rate analysis of micro-organisms in petri dishes. Leveraging frameworks like OpenCV, Mask-RCNN, and PyTorch, I created a robust detection pipeline that increased accuracy from 80% to 94%, significantly outperforming traditional hardware-based solutions. This resulted in substantial cost savings of INR 1,00,000 per hardware unit.

To further enhance efficiency, I implemented an annotation pipeline using Raspberry Pi and a CMOS camera, reducing annotation costs by INR 6,00,000 for 15 micro-organism species. The solution incorporated active learning techniques to minimize manual intervention, streamlining the detection process while maintaining high accuracy.

This project demonstrates my expertise in applying computer vision and machine learning to solve real-world problems, delivering scalable, cost-effective, and highly accurate solutions for life sciences applications.