PyNET-Q×Q: An Efficient PyNET Variant for Q×Q Bayer Pattern Demosaicing in CMOS Image Sensors

Deep learning-based image signal processor (ISP) models for mobile cameras can generate high-quality images that rival those of professional DSLR cameras. However, their computational demands often make them unsuitable settings. Additionally, modern employ non-Bayer color filter arrays (CFA) such as Quad Bayer, Nona and QxQ Bayer to enhance quality, yet most existing deep ISP (or demosaicing) focus primarily on standard CFAs. In this study, we present PyNET-QxQ, a lightweight demosaicing model specifically designed CFA patterns, which is derived from the original PyNET. We also propose knowledge distillation method called progressive train reduced network more effectively. Consequently, PyNET-QxQ contains less than 2.5% parameters PyNET while preserving its performance. Experiments using captured by proto type camera sensor show outperforms conventional algorithms in terms texture edge reconstruction, despite significantly parameter count.