Object Detection Based Drunk and Drive Prevention using Ignition Interlock System

The prevalence of alcohol-related accidents and fatalities on roads necessitates the development of innovative technologies to enhance individual and public safety. In response to this need, our project introduces an advanced system for drunk individual detection and ignition interlock, utilizing a non-invasive approach based on the analysis of pupil dilation. The proposed system aims to provide a reliable and efficient method for classifying and determining individuals as either sober or intoxicated, ultimately influencing the functioning of an ignition interlock device through binary digit representation. Unlike traditional methods such as breathalyzers, our system offers a non-contagious and discrete solution, reducing the likelihood of discomfort or resistance from individuals undergoing the test. The methodology involves capturing and analyzing physiological responses, with a focus on pupil dilation as a well-established biological factor correlated with alcohol intoxication. Analysis of Variance is employed for each individual’s pupil dilation data, forming the basis of a robust statistical model capable of distinguishing between sober and drunk states based on the variance in pupil dilation among individuals. To enhance the system’s accuracy, we leverage a dataset of faces and pupils from individuals under the influence of alcohol. This dataset is crucial for training and fine-tuning the facial recognition and pupil dilation analysis algorithms. By incorporating real-world examples of intoxicated individuals the system