Eyasu Asnake

Hey, I'm the solo developer behind this project. Here's the story: I got frustrated with the traditional remote sensing workflow. If you want to detect "cars," you train a CNN on a cars dataset. Need to find "blue shipping containers" next week? Start over with new data and a new model. The fixed-class bottleneck is brutal. So I built a pipeline that uses Vision-Language Models (VLMs) to skip...