MonoGaze

This project centers on innovative inter-vehicular depth estimation using monocular RGB camera feeds, with a special focus on headway (distance to the vehicle ahead) in urban traffic scenarios like those found in Guwahati City. By harnessing cutting-edge deep learning methods for depth extraction and integrating behavioural analysis-including speed, acceleration, and driver concentration via facial emotion detection-the project provides rich insights into driving patterns in congested conditions. Validated against stereo-camera data, the system utilizes pretrained depth models and car detection (Roboflow), offering actionable data to improve road safety and understand driver responses to environmental stimuli. This foundational work paves the way for future enhancements, including dataset expansion, model accuracy improvements, and broader investigati