Detecting colour vision deficiencies via Webcam-based Eye-tracking: A case study

Attached is WECVD - Webcam-based eye-tracking dataset from CVD (Colour Vision Deficiency) subjects. 12 subjects volunteered to participate in a webcam-based eye-tracking session. Half of the participants are individuals with CVD (Colour Vision Deficiency), while the other half are non-CVD subjects. Fourteen images were prompted to the screen for a time range of 5 seconds each. A one-second transition frame was shown to refresh viewers' retinas and move to the following images. On-screen eye movements of each participant were recorded. All images contain regions having red as the dominant colour. That is because red is a colour dichromatic vision systems such as protanopic and deuteranopic ones struggle to perceive. Those regions are inscribed into rectangular masks and defined as AOI (areas of interest). Differences between CVD and normal vision systems are sought by looking at percentages of fixation points in AOIs for each image within the same group. Intergroup comparison gave the overall description of the case study. The hypothesis of CVD and normal vision system groups showing different paths over the first 3 seconds of observation was tested using one-way ANOVA. Results proved the statistical significance of the test with a p-value < 0.05 Images folder contains 14 RGB images; Heatmaps folder includes CVD (Colour Vision Deficiency) and NVS (Normal Vision System) eye-movement distribution as heatmaps Experiments_GitHub.xlsx is a spreadsheet file with all details about the fixation point number for each image, stats and percentage of AOI (Areas of Interest) points. https://dl.acm.org/doi/10.1145/3588015.3590133#WECVD-main.zip