Model-Based Image Signal Processors via Learnable Dictionaries

Authors: Marcos V. Conde, Steven McDonagh, Matteo Maggioni, Ales Leonardis, Eduardo Pérez-Pellitero481-489

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evidence the value of our data generation process by extensive experiments under both RAW image reconstruction and RAW image denoising tasks, obtaining state-of-the-art performance in both.
Researcher Affiliation Collaboration Huawei Noah s Ark Lab *Marcos V. Conde is now with Computer Vision Lab, Institute of Computer Science, University of W urzburg.
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide a specific link or explicit statement about the public availability of its source code in the main text.
Open Datasets Yes SIDD (Abdelhamed, Lin, and Brown 2018; Abdelhamed, Timofte, and Brown 2019). ... MIT-Adobe Five K dataset (Bychkovsky et al. 2011). We use the train-test sets proposed by Inv ISP (Xing, Qian, and Chen 2021) for the Canon EOS 5D and the Nikon D700, and the same processing using the Lib Raw library to render ground-truth s RGB images from the RAW images.
Dataset Splits Yes There are 320 ultra-high-resolution image pairs available for training (e.g. 5328 3000). Validation set consist of 1280 image pairs.
Hardware Specification No The paper states 'For more details we refer the reader to the supplementary material, where we also provide other relevant information about the training process e.g. GPU devices', but it does not provide specific hardware models (like GPU types, CPU types, or memory) in the main text.
Software Dependencies No The paper mentions 'Lib Raw library' for processing, but it does not specify versions for software dependencies or other key libraries used in the experiments.
Experiment Setup No The paper mentions 'batch sizes, network architectures' are provided in the supplementary material, but it does not provide specific hyperparameter values or detailed training configurations in the main text.