Scholarlink

This study introduces a deep learning solution to safeguard children from potentially harmful content in cartoons, such as violence and adulterous scenes. Recognizing the possible negative influence of such content on young minds, we developed a hybrid Convolutional Neural Network (CNN) model that combines the strengths of InceptionV3 and VGG16. This model is meticulously designed to effectively identify and filter out aggressive and inappropriate content. Our comparative analysis highlights the proposed model’s superior performance, including established models like VGG16, VGG19, ResNet50, and InceptionV3. It is particularly notable for its high efficiency and fewer parameters, successfully addressing the common issue of overfitting often encountered in deep learning models. This advancement in performance is critical for the model's ability to identify and mitigate exposure to harmful content in cartoons accurately. The results of this research represent a significant advancement in content moderation technology for children's media, paving the way for a safer and more suitable viewing environment.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deep Learning in Cartoon Moderation: Distinguishing Child-Friendly Content with CNN Architectures

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • S Kumar Reddy Mallidi,
  • Sujana Bellapukonda,
  • Nikhitha Gollapudi,
  • Karthik Malaka,
  • P. Sreeshanth,
  • Sai Sri Harsha Polisetti

摘要

This study introduces a deep learning solution to safeguard children from potentially harmful content in cartoons, such as violence and adulterous scenes. Recognizing the possible negative influence of such content on young minds, we developed a hybrid Convolutional Neural Network (CNN) model that combines the strengths of InceptionV3 and VGG16. This model is meticulously designed to effectively identify and filter out aggressive and inappropriate content. Our comparative analysis highlights the proposed model’s superior performance, including established models like VGG16, VGG19, ResNet50, and InceptionV3. It is particularly notable for its high efficiency and fewer parameters, successfully addressing the common issue of overfitting often encountered in deep learning models. This advancement in performance is critical for the model's ability to identify and mitigate exposure to harmful content in cartoons accurately. The results of this research represent a significant advancement in content moderation technology for children's media, paving the way for a safer and more suitable viewing environment.