NLP and computer vision were developed as separate fields; however, researchers are now combining them to advance the process of content moderation. Computer vision is now being used to classify images, and optical character recognition is used to extract any text from the image. This text is then processed using NLP to classify the text and detect any offensive, toxic and suggestive content for moderation. The image moderation model used generates probability scores on the likelihood that an image contains violence, gore, suggestive nudity, drugs or explicit nudity, etc.
Download the full whitepaper to learn more.