Scholarlink

One of the most popular leisure activities is watching movies. Hype surrounding it contributes to the collections which are an indicator of the financial success. Genre classification is of paramount importance for movie recommendation systems. Accurate prediction of the genre is difficult owing to the high variations in texture, resolution etc. Our research for genre classification has relied upon the features extracted from posters as well as social media. To solve this multi-label problem, we have proposed a unique approach by joint modeling of the meta-data and data elicited from Facebook, YouTube, etc., that helps predict the verdict at box office. A deep neural network, Convolutional Neural Network is used on a diverse dataset curated for the purpose of genre prediction that achieves impressive accuracy. We have compared our proposed multi-modal model with the current state-of-the-art techniques, and results show that our model significantly outperforms the others in terms of accuracy (96%), for top ten (10) genres.

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

A Multi-label Classification of Movie Genre for Prediction of Box-Office (BO) Verdict Using Posters and Social Media Sarcasm: A Multimodal Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Poulami Dutta,
  • Chandan Kumar Bhattacharyya

摘要

One of the most popular leisure activities is watching movies. Hype surrounding it contributes to the collections which are an indicator of the financial success. Genre classification is of paramount importance for movie recommendation systems. Accurate prediction of the genre is difficult owing to the high variations in texture, resolution etc. Our research for genre classification has relied upon the features extracted from posters as well as social media. To solve this multi-label problem, we have proposed a unique approach by joint modeling of the meta-data and data elicited from Facebook, YouTube, etc., that helps predict the verdict at box office. A deep neural network, Convolutional Neural Network is used on a diverse dataset curated for the purpose of genre prediction that achieves impressive accuracy. We have compared our proposed multi-modal model with the current state-of-the-art techniques, and results show that our model significantly outperforms the others in terms of accuracy (96%), for top ten (10) genres.