Dragonfly is an innovative AI solution that replicates the way the human eye and brain processes visual information to identify which elements in an image or video is grabbing the viewers' attention.
The biological model is based on over 10 years research from world leading scientists at Queen Mary University of London in the field of Visual Perception and mathematical models for visual processing.
The solution is a simple yet powerful iOS application for iPad and iPhone enabling instant analysis and interactive features to surface objective and quantitative metrics behind your creatives.
Dragonfly helps you define, plan and test your campaign creative … in real-time
It provides immediate differentiation to traditional research-based approaches such as eye tracking and panel-based studies. Offering unique insights into human visual attention in the first 1-2 seconds before individual human biases kick in.
Dragonfly is currently only supported on Apple iOS devices
Minimum requirements
Recommended requirements
Unfortunately, not, Dragonfly is currently only supported on Apple iOS devices.
Not at the moment, but we're working on it.
Due to technical and processing limitations it is not currently possible to analyse pre-shot video clips from the iOS camera roll. You can however use the real-time functionality to analyse the live video feed from your devices camera.
The solution is based on over 10 years extensive research carried out by the Vision & Robotics group at Queen Mary University of London into change blindness and visual attention.
The solution analyses and detects which areas on any given static image or real-time video footage captured on your iOS camera is more likely to attract the user's attention first.
Visual Saliency is the state or quality in which visual information stands out relative to its neighbours. This enables us to identify which objects or elements are more likely to automatically attract our attention before others.
The DRAGONFLY solution is based on a biological computational model developed by Queen Mary University of London that replicates human 'visual pathways' to enable us to identify what is most likely to grab our attention - 'visual saliency' within what we see around us.
As human's we all see the same thing in the first few seconds, our unique algorithm analyses each pixel based on 5 key things: Luminance, Contrast, Texture, Scale & Orientation, assigning an average score per pixel. Finally, you can scientifically prove which creative works best to get your attention
Dragonfly provide instant analysis of your image-based content accessed from your devices' camera roll in the iOS Photos app. It also provides real-time heat-map analysis of video footage from your devices camera
Dragonfly's system performance of the AI model has been validated on the MIT 300 Salience Benchmark database - this is used as a resource to enable researchers to benchmark, compare and promote progress in the field of saliency modelling. The Dragonfly AI model achieved a 89.2% prediction efficiency with the eye fixation data.
For more information, please visit MIT Saliency Benchmark
If you have any further questions, please submit a request