Hand-Drawn Electronic Component Detection and Simulation Using Deep Learning and Flask

This paper presents a key innovation - the creation of a highly user-friendly web application that allows users to easily upload their hand-drawn circuit diagrams and instantly receive LTspice-compatible .asc files. This level of automation and simplicity marks a significant breakthrough in electronic design automation. A system is designed to automate the simulation of hand-drawn circuit diagrams, bridging the gap between manual circuit design and simulation readiness. Leveraging a custom convolutional neural network (CNN) model trained on a specialized custom dataset, the system accurately recognizes over 25 distinct electronic components from hand-drawn schematics. The detected components are translated into LTspice Netlist (.asc) files, enabling seamless circuit simulation. The methodology involved understanding the problem, creating the custom dataset, developing the custom CNN model, and integrating the backend with a user-friendly web interface using Flask. The proposed custom deep learning model outperformed existing solutions with a training accuracy of 99% and a test accuracy of 96%. This approach simplifies the traditionally complex simulation process, allowing educators and students to focus on circuit design principles while enhancing accessibility and enriching electronics education.