| Proj No. | A2093-251 |
| Title | Generative Deep Learning for Controllable PCB Image Synthesis |
| Summary | Recent advancements in deep generative models, such as Variational Autoencoders (VAEs) and diffusion-based methods, have transformed natural image synthesis, enabling sophisticated controllable generation techniques like text-to-image synthesis. However, applying these approaches to the PCB domain presents unique challenges. PCB images are highly structured and demand precise control over various design parameters—including trace routing, component placement, and other critical specifications—to meet stringent quality standards. This project aims to adapt and extend these controllable generative techniques to the PCB scenario by incorporating control signals derived from PCB design and manufacturing constraints. Ultimately, this work strives to bridge the gap between natural and PCB image generation, enhancing automated PCB inspection and design verification workflows. |
| Supervisor | A/P Gwee Bah Hwee (Loc:S1 > S1 B1B > S1 B1B 42, Ext: +65 67906861) |
| Co-Supervisor | - |
| RI Co-Supervisor | - |
| Lab | IC DESIGN II (Loc: S1-B2B-10) |
| Single/Group: | Single |
| Area: | Digital Media Processing and Computer Engineering |
| ISP/RI/SMP/SCP?: |