Netlist from Hand-Drawn Circuits
Converting hand-drawn circuits to LTSpice netlists via image processing and ML (AISP 2022)
My undergraduate research converted hand-drawn circuit diagrams into simulation-ready netlists. I built an ML pipeline combining Histogram of Gradients features with an SVM classifier, plus adaptive binarization and skeletonization for component detection — reaching up to 95% accuracy across five component types. An end-point analysis algorithm with OpenCV-based OCR converted 100+ hand-drawn circuits directly into LTSpice-compatible format.