Traditional to Modern Tools: A Comparative Study of Visual Lip Feature Extraction with Dlib

Abstract

Visual Lip refers to the visible movements and contour of the lips during speech. It shows how the mouth moves when pronouncing sounds without any audio. This study report describes the various visual lip feature extraction methods and tools with the steps involved in each tool. Overall Relationship between different lip extraction Models and tools also discussed in study report. The current study reports the status of research literature chronologically, explaining various tools for lip feature extraction mainly focusing set up dependency for Dlib tool. The current study also explains various lip feature extraction tools including dlib, Mediapipe and OpenCV with conducting a comparative study, with its description, speed, accuracy, advantages, disadvantages and output with each tool. Set up dependency with each package and version for Dlib tool as it has version compatibility errors also taken into consideration in this report. Paper reports 10 common Visual Features that extracted from Dlib with its calculation.  Our findings show that lip extraction has a variety of uses in everyday life.  Various Challenges for feature extraction while using Dlib tool are also discussed in the paper. Until now, research on visual lip extraction using each tool has been conducted independently. Based on our findings, we determined that the comparative and in-depth dependency and setup for Dlib tool. Therefore this work is unique and published for the first time.