Doodle Detector
The Project
Doodle Detector is an interactive tool that uses machine learning to recognize hand-drawn doodles and identify what they represent. If you’re unsure about the name of an object or just want to explore recognition technology, you can sketch a simple drawing, and the system will analyze it and tell you what it is—whether it's a circle, a cat, or something more complex. It’s a fun, educational way to learn about object recognition and see AI in action. The idea was inspired by the way children and even adults use drawing as a means of expression and learning. Sometimes we recognize objects visually but can’t recall their names—this tool bridges that gap. Inspired by tools like Quick, Draw! by Google, this project brings the experience into a personal, hands-on space where curiosity leads to discovery. One of the biggest challenges was training the model to accurately recognize rough and varied doodles. Since everyone draws differently, the system had to be flexible and accurate across many drawing styles. Balancing simplicity and performance in the model while keeping the experience fast and responsive was also a key technical hurdle.
