FSID: Fully Synthetic Image Denoising via Procedural Scene Generation
This addresses scalability and privacy issues in low-level vision tasks for commercial applications, though it is incremental as it applies existing methods to new synthetic data.
The paper tackles the problem of training image denoising models without relying on real-world data by developing a procedural synthetic data generation pipeline, generating 175,000 noisy/clean image pairs, and shows that a model trained on this synthetic data achieves competitive denoising results on real smartphone images.
For low-level computer vision and image processing ML tasks, training on large datasets is critical for generalization. However, the standard practice of relying on real-world images primarily from the Internet comes with image quality, scalability, and privacy issues, especially in commercial contexts. To address this, we have developed a procedural synthetic data generation pipeline and dataset tailored to low-level vision tasks. Our Unreal engine-based synthetic data pipeline populates large scenes algorithmically with a combination of random 3D objects, materials, and geometric transformations. Then, we calibrate the camera noise profiles to synthesize the noisy images. From this pipeline, we generated a fully synthetic image denoising dataset (FSID) which consists of 175,000 noisy/clean image pairs. We then trained and validated a CNN-based denoising model, and demonstrated that the model trained on this synthetic data alone can achieve competitive denoising results when evaluated on real-world noisy images captured with smartphone cameras.