Hate Speech and NSFW Image Classification using BERT and ResNet-34 model

As social media and internet platforms have grown in popularity, hate speech has proliferated as well, making many users feel uncomfortable and endangered. This study offers a fresh solution to this problem by successfully classifying hate speech using a combination of text and picture analysis. The suggested multimodal technique makes use of the advantages of two potent machine learning algorithms: image classification using ResNet-34 and text classification using BERT. ResNet- 34 can extract relevant features from photos to discriminate between neutral and hateful visual content, BERT can reliably detect hate speech patterns in text by capturing the context and subtleties of language. This study offers a strong foundation for recognizing and removing harmful information in both text and visual formats, which has important ramifications for upholding a secure and welcoming online community.