Image-based Addiction Detection: An Automated Approach

The emergence of digital imagery and advancements in computer vision techniques have opened new opportunities for addressing public health concerns, including addiction detection. This paper focuses on automated recognition of cigarette smoking behavior in images, leveraging the YOLOv8 architecture. This study compiles a comprehensive dataset of annotated cigarette images and fine-tune the proposed model through iterative training. Experimental results demonstrate high accuracy and recall in cigarette detection, highlighting the efficacy of our approach. This research study offers a robust solution with broad implications for public health surveillance, tobacco control, movie censorship, and regulatory compliance monitoring.