SynthRSF: A Novel Photorealistic Synthetic Dataset for Adverse Weather Condition Denoising
Angelos Kanlis, Vazgken Vanian, Sotiris Karvarsamis, Ioanna Gkika, Konstantinos Konstantoudakis, Dimitrios Zarpalas · 2024
This paper presents the SynthRSF dataset for training and evaluating single-image rain, snow and haze denoising algorithms, as well as evaluating object detection, semantic segmentation, and depth estimation performance in noisy or denoised images. Our dataset features 26,893 noisy images, accompanied by the associated ground truth images. It further includes 13,800 noisy images accompanied by ground truth, 16-bit depth maps and pixel-accurate annotations for various object instances in each frame. The utility of SynthRSF is evaluated by training unified models for rain, snow and haze removal and receiving good objective metrics as well as excellent subjective results in comparison with previously published adverse weather condition datasets. Furthermore, we demonstrate its use as a benchmark for the performance of an object detection algorithm in weather-degraded image datasets.