Technology Detail

Our predictive eye tracking software uses database of eye movements and gaze patterns of actual users

The software analyses people’s eye movement and gaze pattern data such as fixations (steady gaze), saccades (quick movements), and following a moving object is measured and benchmarked This data is used to understand the viewer’s unconscious priorities while viewing any design stimuli’s. We compare and tune the analysis with use of both types of data with public eye tracking data sets and proprietary. The database containing thousands of images viewed by hundreds of participants from real eye tracking studies is then used to train our algorithm. We measure, benchmark, compare and tune the analysis with use of both types of data with public eye tracking data sets and proprietary.


Algorithm Inspired by Neuro science research

Neuroscience helps us to understand how people perceive, process and respond to your design. Dhiti AI simulates human vision and predicts what people will automatically see or miss in first few seconds of viewing. It analyses your designs using artificial neural networks (ANN) algorithm, that simulate what people see during the first 5 seconds of viewing. With image processing and machine learning the model map out the underlying visual perceptual patterns and features in the data like contrast, orientation, layout, size, intensity, colour, text and face detection. Dhiti AI’s attention prediction model is based on deep learning and can automatically detect visual attention shifts that can be used as a substitution for eye tracking studies.



Accuracy

Up to date, we have achieved 90% accuracy in our datasets containing data of how humans look at the designs for the first 5 seconds of viewing. State of the art accuracy in Saliency prediction, as known as Attention (Heat) maps. Dhit AI’s predictions to be 90% as accurate as conducting physical studies involving hundreds of participants, eye tracking hardware, lot of money and weeks of time. Simplified everything with just a click. It can analyse almost any visual like print and digital advertisements, indoor/outdoor signage, websites, apps, e-mail, packaging & shelf layouts