Realistic Bokeh Effect Rendering on Mobile GPUs, Mobile AI & AIM 2022 challenge: Report
Authors:
- Andrey IgnatovAIAndrey Ignatov
- Radu TimofteRadu Timofte
- Jin ZhangJZJin Zhang
- Feng ZhangFZFeng Zhang
- Gaocheng YuGYGaocheng Yu
- Zhe MaZMZhe Ma
- Hongbin WangHWHongbin Wang
- Minsu KwonMKMinsu Kwon
- Haotian QianHQHaotian Qian
- Wentao TongWTWentao Tong
- Pan MuPMPan Mu
- Ziping WangZWZiping Wang
Preprint
Abstract
As mobile cameras with compact optics are unable to produce a strong bokeh effect, lots of interest is now devoted to deep learning-based solutions for this task. In this Mobile AI challenge, the target was to develop an efficient end-to-end AI-based bokeh effect rendering approach that can run on modern smartphone GPUs using TensorFlow Lite. The participants were provided with a large-scale EBB! bokeh dataset consisting of 5K shallow / wide depth-of-field image pairs captured using the Canon 7D DSLR camera. The runtime of the resulting models was evaluated on the Kirin 9000's Mali GPU that provides excellent acceleration results for the majority of common deep learning ops. A detailed description of all models developed in this challenge is provided in this paper.Keywords
All Related Versions
This publication has 12 references indexed in Scilit:
- Reversed Image Signal Processing and RAW Reconstruction. AIM 2022 Challenge Report2023
- Simple Baselines for Image Restoration2022
- Constrained Predictive Filters for Single Image Bokeh RenderingIEEE Transactions on Computational Imaging, 2022
- Depth-aware blending of smoothed images for Bokeh effect generationJournal of Visual Communication and Image Representation, 2021
- Controlling information capacity of binary neural networkPattern Recognition Letters, 2020
- AIM 2020 Challenge on Rendering Realistic Bokeh2020
- AIM 2020 Challenge on Learned Image Signal Processing Pipeline2020
- AIM 2019 Challenge on RAW to RGB Mapping: Methods and Results2019
- Deep Automatic Portrait Matting2016
- Perceptual Losses for Real-Time Style Transfer and Super-Resolution2016