Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

PixLift: Accelerating Web Browsing via AI Upscaling

Yonas Atinafu, Sarthak Malla, HyunSeok Daniel Jang, Nouar Aldahoul, Matteo Varvello, Yasir Zaki

TL;DR

The paper addresses data-limited internet access in developing regions by targeting image-sized contributions to webpage data. It proposes PixLift, a system that downscales images during transmission and upscales them on-device using AI, implemented as a Chromium extension with a TFJS-based on-device super-resolution pipeline. Key contributions include a feasibility analysis of remote image downscaling, a practical browser-extension design, and comparative evaluations of three super-resolution models, highlighting QuickSRNet Small 4X as the most time-efficient option with acceptable quality trade-offs. The findings show significant data savings and median page-load-time improvements under full support, suggesting PixLift can meaningfully reduce costs and improve accessibility in bandwidth-constrained environments. The work lays groundwork for adaptive model selection and broader deployment to promote more inclusive web access.

Abstract

Accessing the internet in regions with expensive data plans and limited connectivity poses significant challenges, restricting information access and economic growth. Images, as a major contributor to webpage sizes, exacerbate this issue, despite advances in compression formats like WebP and AVIF. The continued growth of complex and curated web content, coupled with suboptimal optimization practices in many regions, has prevented meaningful reductions in web page sizes. This paper introduces PixLift, a novel solution to reduce webpage sizes by downscaling their images during transmission and leveraging AI models on user devices to upscale them. By trading computational resources for bandwidth, PixLift enables more affordable and inclusive web access. We address key challenges, including the feasibility of scaled image requests on popular websites, the implementation of PixLift as a browser extension, and its impact on user experience. Through the analysis of 71.4k webpages, evaluations of three mainstream upscaling models, and a user study, we demonstrate PixLift's ability to significantly reduce data usage without compromising image quality, fostering a more equitable internet.

PixLift: Accelerating Web Browsing via AI Upscaling

TL;DR

The paper addresses data-limited internet access in developing regions by targeting image-sized contributions to webpage data. It proposes PixLift, a system that downscales images during transmission and upscales them on-device using AI, implemented as a Chromium extension with a TFJS-based on-device super-resolution pipeline. Key contributions include a feasibility analysis of remote image downscaling, a practical browser-extension design, and comparative evaluations of three super-resolution models, highlighting QuickSRNet Small 4X as the most time-efficient option with acceptable quality trade-offs. The findings show significant data savings and median page-load-time improvements under full support, suggesting PixLift can meaningfully reduce costs and improve accessibility in bandwidth-constrained environments. The work lays groundwork for adaptive model selection and broader deployment to promote more inclusive web access.

Abstract

Accessing the internet in regions with expensive data plans and limited connectivity poses significant challenges, restricting information access and economic growth. Images, as a major contributor to webpage sizes, exacerbate this issue, despite advances in compression formats like WebP and AVIF. The continued growth of complex and curated web content, coupled with suboptimal optimization practices in many regions, has prevented meaningful reductions in web page sizes. This paper introduces PixLift, a novel solution to reduce webpage sizes by downscaling their images during transmission and leveraging AI models on user devices to upscale them. By trading computational resources for bandwidth, PixLift enables more affordable and inclusive web access. We address key challenges, including the feasibility of scaled image requests on popular websites, the implementation of PixLift as a browser extension, and its impact on user experience. Through the analysis of 71.4k webpages, evaluations of three mainstream upscaling models, and a user study, we demonstrate PixLift's ability to significantly reduce data usage without compromising image quality, fostering a more equitable internet.

Paper Structure

This paper contains 8 sections, 2 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Image Scaling Support
  4. PixLift Design
  5. PixLift Evaluation
  6. Methodology
  7. Results
  8. Conclusion

Figures (2)

  • Figure 1: PixLift performance evaluation: resource usage and visual quality assessment via user study.
  • Figure 2: PixLift performance evaluation in the wild (1,000 most popular Pakistani webpages). Today refers to current support for remote image downscaling; full refers to assuming all images can be downscaled.