A custom convolutional neural network capable of detecting 10 hand-drawn circuit components with a classification accuracy of 94.94 percent is presented in this project. The model features a simple design that may be used on devices with limited computing power.
The "CktComponentRecognizer.py" file consists of the neural network architecture, programmed using the Pytorch ML Framework in Python 3.8.
Original images acquired for the dataset: https://drive.google.com/drive/folders/14J6al19sL4wZ3UkMM_SYQ_00xxmEDO6r?usp=sharing
A simple GUI has been developed using Tkinter and Pygame libraries for demonstration purposes.
Paper drafted for this project has also been attached.
NOTE: Make sure all the code files and dataset folders are in the same directory for a error-free execution.