Quantized Image Super-Resolution on Mobile Npus, Mobile AI 2025 Challenge: Report
Andrey Ignatov, Georgy Perevozchikov, Radu Timofte, Zhiyu Zhang, Tianxiao Gao, Yukun Yang, Shiai Zhu, Shihao Wang, Kihwan Yoon, Ganzorig Gankhuyag, Hyeon-Cheol Moon, Taehyun Jeong, Yu‐Mi Kim, Suhyeon Lee, Jaehun Baek, Jin-Woo Jeong, Eunjun Park, Jun Haeng Lee, Hee-Jun Lee, Sungjei Kim, Dafeng Zhang, Yong Yang, Heo Myeong Cheol, Yonghyun Park, J. H. Jeong, Wontae Kim, Kanghwan Lee, Diankai Zhang, Biao Wu, Chengjian Zheng, Shaoli Liu, Si Gao, Ning Wang, Mingshen Wang, Zhao Zhang, Suiyi Zhao, Jinhan Guan, Bo Wang, Yan Luo
Abstract
Image super-resolution is a classic computer vision problem with numerous practical applications on mobile and IoT devices. This creates a need for solutions that are not only performant but are additionally compatible with real mobile AI hardware such as neural processing units (NPUs). In this Mobile AI challenge, we address this problem and propose the participants to design efficient quantized deep learning super-resolution models that can demonstrate a real-time performance on mobile NPUs. For this, the participants were provided with the DIV2K dataset and trained quantized models to do an efficient <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$3 X$</tex> image upscaling. The runtime of all models was evaluated on the Google Tensor NPU present in all recent Google Pixel smartphones. The proposed solutions are fully compatible with all major mobile AI accelerators and are capable of reconstructing Full HD images under 50 ms, delivering highfidelity results. A comprehensive description of the models developed in the challenge is provided in this paper.