The Science Behind the Platform

Attention is
measurable
before launch

VeysaLabs uses peer-reviewed saliency science to predict where real human attention lands: before a campaign launches. No guesswork. No post-mortems. Just evidence.


The Problem

Creative decisions
have always been made blind

Every day, marketing teams and agencies make high-stakes decisions about visual creative: which headline leads, where the product sits, how much space a logo commands: based on instinct, hierarchy, or whoever spoke loudest in the briefing.

Traditional eye tracking offers a solution, but at a cost: lab recruitment, specialist equipment, weeks of lead time, and budgets that only enterprise brands can justify. By the time results come back, the creative has already gone live.

The science of visual attention has existed for decades. VeysaLabs makes it accessible at the speed of production.

70ms
Time for the human visual system to orient initial attention to a scene
3–5
Fixations a viewer makes in the first two seconds of seeing an ad
<1s
Time for VeysaLabs to generate a full attention prediction on your creative

How It Works

From pixels to
attention prediction

VeysaLabs applies a multi-stage computational pipeline to predict where viewers will look: grounded in two decades of attention research and a model trained on large-scale human fixation data.

01

Upload a creative or enter a URL

Two ways in. Upload an image directly: a banner, social post, OOH, print ad, or digital asset. Or paste any live webpage URL and VeysaLabs automatically captures a full-page screenshot for analysis. Standard formats accepted at production resolution. Your creative is processed securely and never used to train or update models.

Image upload URL scan
02

Transformer-enhanced saliency prediction

Your image is passed through TranSalNet: a state-of-the-art saliency model that combines convolutional neural networks with transformer encoders. Transformers capture long-range spatial dependencies in the image that CNN-only architectures miss, producing a more perceptually accurate attention map.

The model was trained on the SALICON dataset: 10,000 images annotated with human fixation data collected via a validated psychophysical paradigm. It achieves top-tier performance on the MIT300 benchmark: the standard evaluation for saliency prediction models.

TranSalNet · Lou et al., Neurocomputing 2022
03

Attention map generation

The model outputs a continuous saliency map: a spatial probability distribution showing where attention is likely to concentrate. High-density regions indicate where viewers will fixate first and longest. The map is rendered as a heatmap overlay on your original creative.

Analysis
04

Creative Effectiveness Report

Attention data is interpreted against a framework grounded in published advertising attention research: covering where the eye lands relative to brand elements, product, text hierarchy, and call to action. You receive a structured report with specific, actionable findings.

Output

Research Foundations

Built on 20 years
of attention science

The interpretation framework in every VeysaLabs report is grounded in peer-reviewed research: not internal heuristics. These are the foundational studies that define how visual attention operates in advertising contexts.

2004 · Journal of Marketing Research

Split-second sequential processing: Visual attention in advertising

Established that viewers allocate attention sequentially across ad elements, with brand and pictorial elements receiving primary fixations. The foundational model for understanding attention hierarchy in advertising.

Pieters, R. & Wedel, M. (2004)
2008 · Journal of Consumer Research

Attention capture and transfer in advertising

Identified how size, colour, and informational content interact to determine which ad elements capture and hold viewer attention. Directly informs how we score attention distribution across creative elements.

Wedel, M. & Pieters, R. (2008)
2022 · Neurocomputing

TranSalNet: Towards perceptually relevant visual saliency prediction

The architecture underlying VeysaLabs' inference engine. Integrates transformer encoders into a CNN-based saliency model, achieving state-of-the-art performance on the MIT300 benchmark and the SALICON Saliency Prediction Challenge.

Lou, J., Lin, H., Marshall, D., Saupe, D., & Liu, H. (2022)
2023 · IEEE Transactions

Audience-level visual attention and advertising effectiveness

Recent work bridging computational saliency prediction with real-world advertising outcomes: demonstrating the validity of model-based attention prediction as a proxy for human fixation behaviour in applied marketing contexts.

Xie, et al. (2023)

A note on accuracy, and why we're transparent about it

Synthetic eye tracking predicts where attention is likely to go based on patterns in how humans process visual scenes. It is not a perfect replica of individual eye tracking sessions, and we don't claim otherwise.

What it is: a fast, scalable, research-validated signal that has been shown to correlate strongly with real fixation data at the population level. For creative teams making decisions at speed and scale, it is a material upgrade over instinct: and orders of magnitude faster and cheaper than lab-based testing.

We believe honest tools earn more trust than overclaimed ones.


Built on open research

VeysaLabs is built on openly licensed academic work. We are grateful to the researchers and institutions whose contributions make this platform possible.

MIT License

TranSalNet

Saliency prediction model. Copyright © 2022 Jianxun Lou. Used under the MIT License.

Lou, J., Lin, H., Marshall, D., Saupe, D., & Liu, H. (2022). TranSalNet. Neurocomputing, 494, 455–467. DOI: 10.1016/j.neucom.2022.04.080

github.com/LJOVO/TranSalNet ↗
CC BY 4.0

SALICON Dataset

Human fixation annotations used to train the saliency model. Annotations © VIP Lab, University of Minnesota. Licensed under Creative Commons Attribution 4.0.

salicon.net ↗
CC BY 4.0

TranSalNet Paper

The underlying research is published Open Access in Neurocomputing (2022) under Creative Commons Attribution 4.0. Architecture and methodology by Lou et al.

View paper ↗

Where Teams Use It

Attention intelligence
across every environment

Wherever humans look at visual content, the same attention science applies. VeysaLabs is built for agencies today: and expanding into every context where visual decisions matter.

Available now

Agencies & Creative Teams

Validate campaign creative before it reaches the client. Upload an image or paste a live URL and VeysaLabs handles the rest. Test multiple executions fast, present with evidence, and reduce revision cycles driven by subjective feedback.

  • Digital & social advertising
  • OOH and print
  • Brand & packaging design
  • Landing pages via URL scan
  • Live websites & digital campaigns
Coming soon

Retail & In-Store

Predict attention on shelf layouts, point-of-sale materials, and in-store signage before committing to print runs or fixture builds. Know what shoppers will notice: and what they won't.

  • Shelf & planogram design
  • POS and promotional materials
  • Packaging hierarchy
  • Window & display graphics
Coming soon

Events & Exhibition

Optimise stand builds, exhibition graphics, and wayfinding before fabrication. In high-traffic, competitive environments, attention is the only currency that matters.

  • Exhibition stand layouts
  • Event signage & wayfinding
  • Stage & backdrop design
  • Sponsorship asset placement