Impact Factor

7.883 Call For Paper

Share on

Author(s):

G. Bala Murugan | R. Govintha Kishore
Keywords:

Face Identification, Object Detection, Two-Stage Detector, R-CNN, Deep Learning, Feature Extraction, Movie Footage Analysis.
Abstract:
Face Identification And Object Detection In Movie Footage Present Significant Challenges Owing To The Dynamic Nature Of Video Content, Complex Scene Transitions, Rapid Movements, And Occlusions. Existing Systems That Apply Image-based Algorithms Frame-by-frame Are Computationally Intensive, Lack Temporal Consistency, And Fail To Exploit Contextual Information Across Consecutive Frames. This Paper Proposes A Two-stage Deep Learning Pipeline For Face Identification And Object Detection In Movie Footage. The First Stage Employs A Region Proposal Network To Identify Candidate Regions With High Recall, While The Second Stage Applies A Deep Learning-based Classifier To Assign Categorical Labels And Refine Bounding Box Localization. The System Design Is Presented Through Data Flow And Database Design Diagrams, Demonstrating A Structured Pipeline From Image Capture Through Face Detection, Alignment, Feature Extraction, To Final Feature Matching And Application Control.