Smart Bacterial Colony Classification System using visual Images and AI

This project introduces an AI-based system for real-time bacterial colony classification using high-resolution imaging, replacing slow, expert-dependent lab diagnostics. In remote areas of Pakistan, where diagnostic delays are common due to the unavailability of diagnostic labs, doctors often prescribe antibiotics without confirmed test results. Our system analyzes high-resolution 16MP images of 4 colony types across different agar media, resulting in 10 common colony classes. A hybrid model (EfficientNetB3 + ResNet50), trained on 2,200 high-resolution, fresh, and structured images, achieves a combined 95% accuracy. The system is deployed on the cloud via web and mobile apps, offering a fast, low-cost, and expert-free diagnostic solution that aligns with UN SDGs 3 and 9.

Project Poster