Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

NTIRE 2024 Challenge on Night Photography Rendering

Egor Ershov, Artyom Panshin, Oleg Karasev, Sergey Korchagin, Shepelev Lev, Alexandr Startsev, Daniil Vladimirov, Ekaterina Zaychenkova, Nikola Banić, Dmitrii Iarchuk, Maria Efimova, Radu Timofte, Arseniy Terekhin, Shuwei Yue, Yuyang Liu, Minchen Wei, Lu Xu, Chao Zhang, Yasi Wang, Furkan Kınlı, Doğa Yılmaz, Barış Özcan, Furkan Kıraç, Shuai Liu, Jingyuan Xiao, Chaoyu Feng, Hao Wang, Guangqi Shao, Yuqian Zhang, Yibin Huang, Wei Luo, Liming Wang, Xiaotao Wang, Lei Lei, Simone Zini, Claudio Rota, Marco Buzzelli, Simone Bianco, Raimondo Schettini, Jin Guo, Tianli Liu, Mohao Wu, Ben Shao, Qirui Yang, Xianghui Li, Qihua Cheng, Fangpu Zhang, Zhiqiang Xu, Jingyu Yang, Huanjing Yue

TL;DR

The NTIRE 2024 night photography rendering paper surveys a challenge aimed at converting mobile-captured RAW night images into visually pleasing sRGB outputs, while incorporating run-time as a ranking factor alongside perceptual quality. It describes data collected from Huawei Mate 40 Pro devices, three 125-image validation sets, and a 200-image development set, with MOS evaluation via pairwise human judgments and a fixed hardware speed test. The results reveal a spectrum of ISP- and DL-based pipelines, with top methods often building on prior year approaches and achieving a favorable balance between image quality and processing speed, signaling a shift toward efficient deep learning–assisted night imaging on mobile devices. Overall, the work highlights progress in robust, on-device night image processing, emphasizes subjective assessment as a crucial metric, and reinforces openness through reproducible data and methods, advancing practical night photography rendering.

Abstract

This paper presents a review of the NTIRE 2024 challenge on night photography rendering. The goal of the challenge was to find solutions that process raw camera images taken in nighttime conditions, and thereby produce a photo-quality output images in the standard RGB (sRGB) space. Unlike the previous year's competition, the challenge images were collected with a mobile phone and the speed of algorithms was also measured alongside the quality of their output. To evaluate the results, a sufficient number of viewers were asked to assess the visual quality of the proposed solutions, considering the subjective nature of the task. There were 2 nominations: quality and efficiency. Top 5 solutions in terms of output quality were sorted by evaluation time (see Fig. 1). The top ranking participants' solutions effectively represent the state-of-the-art in nighttime photography rendering. More results can be found at https://nightimaging.org.

NTIRE 2024 Challenge on Night Photography Rendering

TL;DR

The NTIRE 2024 night photography rendering paper surveys a challenge aimed at converting mobile-captured RAW night images into visually pleasing sRGB outputs, while incorporating run-time as a ranking factor alongside perceptual quality. It describes data collected from Huawei Mate 40 Pro devices, three 125-image validation sets, and a 200-image development set, with MOS evaluation via pairwise human judgments and a fixed hardware speed test. The results reveal a spectrum of ISP- and DL-based pipelines, with top methods often building on prior year approaches and achieving a favorable balance between image quality and processing speed, signaling a shift toward efficient deep learning–assisted night imaging on mobile devices. Overall, the work highlights progress in robust, on-device night image processing, emphasizes subjective assessment as a crucial metric, and reinforces openness through reproducible data and methods, advancing practical night photography rendering.

Abstract

This paper presents a review of the NTIRE 2024 challenge on night photography rendering. The goal of the challenge was to find solutions that process raw camera images taken in nighttime conditions, and thereby produce a photo-quality output images in the standard RGB (sRGB) space. Unlike the previous year's competition, the challenge images were collected with a mobile phone and the speed of algorithms was also measured alongside the quality of their output. To evaluate the results, a sufficient number of viewers were asked to assess the visual quality of the proposed solutions, considering the subjective nature of the task. There were 2 nominations: quality and efficiency. Top 5 solutions in terms of output quality were sorted by evaluation time (see Fig. 1). The top ranking participants' solutions effectively represent the state-of-the-art in nighttime photography rendering. More results can be found at https://nightimaging.org.
Paper Structure (18 sections, 10 figures, 2 tables)

This paper contains 18 sections, 10 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Challenge
  3. Challenge Data
  4. Evaluation
  5. Results
  6. People's Choice and Discussion
  7. Teams' solutions
  8. Baseline
  9. MiAlgo
  10. SCBC
  11. IVLTeam
  12. DH-AISP
  13. IIR-Lab
  14. PolyuColor
  15. OzU-VGL
  16. ...and 3 more sections

Figures (10)

  • Figure 1: Illustration of final leader board scores and speed of algorithms. Red line separates top-5 quality solutions that were rearranged by inference speed. Winner of efficiency challenge is colored green, team that provided solution with best quality is colored yellow.
  • Figure 2: Example of the impact of high noise level and (color) vignetting on rendering a nighttime image when using only the baseline pipeline.
  • Figure 3: One scene from the final validation set and all results provided by participants. The best quality image is in bold.
  • Figure 4: Overall pipeline of MiAlgo team.
  • Figure 5: SCBC team pipeline scheme.
  • ...and 5 more figures