Skip to content
Switch branches/tags
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time


EyeQ 1.0 overview and a few key insights:

Our paper summarizing key results from EyeQ 1.0 study was accepted into 4th IEEE International Workshop on Mobile Multimedia Computing (MMC 2017). URL for the paper with analysis/insights:

EyeQ / Latest Updates:

EyeQ 2.0 is under active design phase.

  • EyeQ 2.0 includes both no-reference and full-reference modes. Full-reference mode means that both the original and distorted images are available. No-reference mode means that only the distorted images are available.

  • EyeQ 2.0 includes more images especially text-based images.

  • EyeQ 2.0 tracks more information, such as time-to-respond, image display size on the client side, device information, etc.

  • EyeQ 2.0 uses graphicsmagick for progressive compression. URL for graphicsmagick:

What is EyeQ?:

Images are a huge factor in web user experience. Human perception of image quality is known to be subjective and varies with factors such as devices and image contents. Even though image quality assessment (IQA) is a well-researched field, current state-of-the-art algorithms are not built for solving web image delivery problem because they do not consider device variations, image compression artifacts specific to web delivery pipeline, and certain types of images commonly seen on modern web such as text-overlay images. Most importantly, traditional IQA metrics are used to measure image similarity, while in the context of web image delivery, finding the right balance between perceptual quality and image file sizes is a more relevant problem. EyeQ is a crowd-sourcing web application aiming to collect data that would help study web image delivery problem. EyeQ works in Chrome, Firefox and Safari on both mobile and PC platforms. Use of various devices is encouraged.

EyeQ / Goal:

We hope to create an open-source benchmark for assessing image quality that can help improve image delivery on modern web. Our data is going to help researchers understand factors associated with web image delivery that might affect image perceptual quality for end-users.

EyeQ / Team:

Jiawen Zhou - Data Scientist @ Instart Logic
Parvez Ahammad - Head of Data Science & Machine Learning @ Instart Logic
Prasenjit Dey - Software Engineer @ Instart Logic


EyeQ (optimal rate-distortion trade-offs of images on the WWW): Benchmark data, exploratory analyses and insights from large-scale crowd sourcing




No releases published


No packages published