mlsanityData to 7 Oct 2026

NAWQ-SR: A Hybrid-Precision NPU Engine for Efficient On-Device Super-Resolution

arXiv 2212.09501 ↗First posted 15 Dec 2022cs.CVcited by 1

Not on the map. Why?

Abstract (arXiv)

In recent years, image and video delivery systems have begun integrating deep learning super-resolution (SR) approaches, leveraging their unprecedented visual enhancement capabilities while reducing reliance on networking conditions. Nevertheless, deploying these solutions on mobile devices still remains an active challenge as SR models are excessively demanding with respect to workload and memory footprint. Despite recent progress on on-device SR frameworks, existing systems either penalize visual quality, lead to excessive energy consumption or make inefficient use of the available resources. This work presents NAWQ-SR, a novel framework for the efficient on-device execution of SR models. Through a novel hybrid-precision quantization technique and a runtime neural image codec, NAWQ-SR exploits the multi-precision capabilities of modern mobile NPUs in order to minimize latency, while meeting user-specified quality constraints. Moreover, NAWQ-SR selectively adapts the arithmetic precision at run time to equip the SR DNN's layers with wider representational power, improving visual quality beyond what was previously possible on NPUs. Altogether, NAWQ-SR achieves an average speedup of 7.9x, 3x and 1.91x over the state-of-the-art on-device SR systems that use heterogeneous processors (MobiSR), CPU (SplitSR) and NPU (XLSR), respectively. Furthermore, NAWQ-SR delivers an average of 3.2x speedup and 0.39 dB higher PSNR over status-quo INT8 NPU designs, but most importantly mitigates the negative effects of quantization on visual quality, setting a new state-of-the-art in the attainable quality of NPU-based SR.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 38 of its 87 references that are in the map. First posted on arXiv 15 Dec 2022.

Builds on

38 of 87 references.

References · 38

  1. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,827
  2. 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,548
  3. 2017Enhanced Deep Residual Networks for Single Image Super-Resolutioncited by 1,543
  4. 2018Image Super-Resolution Using Very Deep Residual Channel Attention Networkscited by 1,256
  5. 2018ESRGAN: Enhanced Super-Resolution Generative Adversarial Networkscited by 1,250
  6. 2014Image Super-Resolution Using Deep Convolutional Networkscited by 1,218
  7. 2016Real-Time Single Image and Video Super-Resolution Using an Efficient Sub-Pixel Convolutional Neural Networkcited by 1,199
  8. 2017ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devicescited by 1,194
  9. 2018ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Designcited by 1,132
  10. 2017The Perception-Distortion Tradeoffcited by 582
  11. 2018HAQ: Hardware-Aware Automated Quantization with Mixed Precisioncited by 426
  12. 2019HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precisioncited by 387
26 more
  1. 2018Fast, Accurate, and Lightweight Super-Resolution with Cascading Residual Networkcited by 263
  2. 2020Deep Unfolding Network for Image Super-Resolutioncited by 253
  3. 2019HAWQ-V2: Hessian Aware trace-Weighted Quantization of Neural Networkscited by 250
  4. 2019Lightweight Image Super-Resolution with Information Multi-distillation Networkcited by 227
  5. 2020Unfolding the Alternating Optimization for Blind Super Resolutioncited by 129
  6. 2018Fast and Accurate Single Image Super-Resolution via Information Distillation Networkcited by 121
  7. 2019AI Benchmark: All About Deep Learning on Smartphones in 2019cited by 76
  8. 2019Fast, Accurate and Lightweight Super-Resolution with Neural Architecture Searchcited by 66
  9. 2020Rethinking Data Augmentation for Image Super-resolution: A Comprehensive Analysis and a New Strategycited by 66
  10. 2020FracBits: Mixed Precision Quantization via Fractional Bit-Widthscited by 53
  11. 2020PAMS: Quantized Super-Resolution via Parameterized Max Scalecited by 46
  12. 2019Efficient Residual Dense Block Search for Image Super-Resolutioncited by 37
  13. 2021Real-Time Quantized Image Super-Resolution on Mobile NPUs, Mobile AI 2021 Challenge: Reportcited by 34
  14. 2021Extremely Lightweight Quantization Robust Real-Time Single-Image Super Resolution for Mobile Devicescited by 30
  15. 2018Efficient Super Resolution Using Binarized Neural Networkcited by 29
  16. 2019MobiSR: Efficient On-Device Super-Resolution through Heterogeneous Mobile Processorscited by 27
  17. 2020Journey Towards Tiny Perceptual Super-Resolutioncited by 27
  18. 2022SDQ: Stochastic Differentiable Quantization with Mixed Precisioncited by 27
  19. 2021Smart at what cost? Characterising Mobile Deep Neural Networks in the wildcited by 25
  20. 2019EmBench: Quantifying Performance Variations of Deep Neural Networks across Modern Commodity Devicescited by 21
  21. 2019Neural Network Inference on Mobile SoCscited by 17
  22. 2021Deep Neural Network-based Enhancement for Image and Video Streaming Systems: A Survey and Future Directionscited by 9
  23. 2021OODIn: An Optimised On-Device Inference Framework for Heterogeneous Mobile Devicescited by 7
  24. 2020FastVA: Deep Learning Video Analytics Through Edge Processing and NPU in Mobilecited by 6
  25. 2020Neural Enhancement in Content Delivery Systems: The State-of-the-Art and Future Directionscited by 2
  26. 2017clcNet: Improving the Efficiency of Convolutional Neural Network using Channel Local Convolutionscited by 1

Built on by

Cited by year

Citing papers per year: 2024: 1.

1 paper.

  1. 2024Dynamic DNNs and Runtime Management for Efficient Inference on Mobile/Embedded Devicescited by 0