Overexposure Mask Fusion: Generalizable Reverse ISP Multi-Step Refinement
This work addresses the lack of available RAW data for computational photography tasks, enabling enhancements in the RAW domain, though it appears incremental as it builds on existing high-performance methodologies.
The paper tackles the problem of reconstructing RAW sensor data from RGB images, a reverse ISP task, by proposing a multi-step refinement method that integrates an overexposure mask, achieving state-of-the-art performance with significant improvements over baseline U-Net models.
With the advent of deep learning methods replacing the ISP in transforming sensor RAW readings into RGB images, numerous methodologies solidified into real-life applications. Equally potent is the task of inverting this process which will have applications in enhancing computational photography tasks that are conducted in the RAW domain, addressing lack of available RAW data while reaping from the benefits of performing tasks directly on sensor readings. This paper's proposed methodology is a state-of-the-art solution to the task of RAW reconstruction, and the multi-step refinement process integrating an overexposure mask is novel in three ways: instead of from RGB to bayer, the pipeline trains from RGB to demosaiced RAW allowing use of perceptual loss functions; the multi-step processes has greatly enhanced the performance of the baseline U-Net from start to end; the pipeline is a generalizable process of refinement that can enhance other high performance methodologies that support end-to-end learning.