Advancing NSFW Detection in AI: Training Models to Detect Drawings, Animations, and Assess Degrees of Sexiness

摘要

This research explores the advancement of NSFW (Not Safe for Work) detection in AI by training models to detect NSFW content in drawings, animations, and assess degrees of sexiness. Leveraging the NSFWJS library as a foundation, we conduct a comprehensive investigation into enhancing the capabilities of existing NSFW detection models. Through a systematic approach encompassing data collection, annotation, model training, and evaluation, we fine-tune the NSFWJS model to effectively identify NSFW content across diverse media types. Our research addresses the growing need for robust NSFW detection in AI applications, particularly in scenarios involving non-photographic content and nuanced assessments of sexual content. By expanding the capabilities of NSFW detection models, this work contributes to creating safer online environments and enabling more responsible content moderation practices.