Power Efficient Video Super-Resolution on Mobile NPUs with Deep Learning, Mobile AI & AIM 2022 challenge: Report

TL;DR

This paper reports on a challenge to develop real-time, power-efficient video super-resolution models optimized for mobile NPUs, achieving high FPS and low power consumption on a specific mobile platform.

Contribution

The paper introduces a set of optimized deep learning models for mobile video super-resolution that balance high performance with low energy use, tailored for mobile NPUs.

Findings

Models achieve up to 500 FPS on target hardware.

Power consumption is as low as 0.2 Watts at 30 FPS.

All solutions are compatible with the MediaTek Dimensity 9000 NPU.

Abstract

Video super-resolution is one of the most popular tasks on mobile devices, being widely used for an automatic improvement of low-bitrate and low-resolution video streams. While numerous solutions have been proposed for this problem, they are usually quite computationally demanding, demonstrating low FPS rates and power efficiency on mobile devices. In this Mobile AI challenge, we address this problem and propose the participants to design an end-to-end real-time video super-resolution solution for mobile NPUs optimized for low energy consumption. The participants were provided with the REDS training dataset containing video sequences for a 4X video upscaling task. The runtime and power efficiency of all models was evaluated on the powerful MediaTek Dimensity 9000 platform with a dedicated AI processing unit capable of accelerating floating-point and quantized neural networks. All proposed solutions are fully compatible with the above NPU, demonstrating an up to 500 FPS rate and 0.2 [Watt / 30 FPS] power consumption. A detailed description of all models developed in the challenge is provided in this paper.