Diffusion-based image synthesis for domain adaptive IC layout extraction
Abstract
This thesis proposes a diffusion-based framework for conditional Integrated Circuit (IC) image generation and its application to segmentation tasks. To address the limited availability of annotated data and domain differences in IC image analysis, a mask-guided diffusion model is used to generate structurally consistent IC images. A synthetic-only training strategy is adopted, where the segmentation model is trained using synthetic image–mask pairs and evaluated on real IC images. To evaluate the proposed method, three training settings are considered, including a baseline with limited real data and one with traditional data augmentation. The results show that the method can generate IC images with good structural consistency and realistic appearance. The segmentation model trained on synthetic data performs well on real IC images. Compared with the data-scarce baseline, it improves mIoU by 38.2%, Dice Score by 27.7%, and IoU by 9.6%. It also outperforms the traditional augmentation baseline, with improvements of 18.0% in mIoU, 13.2% in Dice Score, and 4.5% in IoU. These results indicate that the method can capture IC structural characteristics and generalize well to real data. Overall, this work shows that diffusion-based image generation can reduce the impact of limited data and improve segmentation performance in IC layout extraction tasks.
Read the paper
Related studies
- Learning from First-Generation Qualitative Approaches in the IS Discipline: An Evolutionary View and Some Implications for Authors and Evaluators (PART 1/2)
- Commentary on the clinical and preclinical dosage limits of interstitially administered magnetic fluids for therapeutic hyperthermia based on current practice and efficacy models
- Prosthesis: New Technological Opportunities and Innovative Biomedical Devices
- The history and future of neural modeling for cochlear implants
- Ryanodine receptors: waking up from refractoriness
- Novel Transgenic Rabbit Model Sheds Light on the Puzzling Role of Matrix Metalloproteinase-12 in Atherosclerosis
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.