Optimizing Reverse Image Search by Generating and Assigning Suitable Captions to Images

Abstract

Reverse image search is a content-based image retrieval (CBIR) query technique that takes a sample image as an input, and search is performed based on it. Reverse image search is characterized by a lack of search terms. Using reverse image search, one can find the original source of images, find plagiarized photos, detect fake accounts on social media, etc. Reverse image search works by uploading an image by the user, and searching of images is carried out by using the corresponding meta tags, HTML tags or color distributions of the image. The search engine currently functions by using this information and not the context of the image, thus resulting in inaccurate outcomes. Labeling the image using keywords relevant to the image is a very generic explanation to an image and can lead to ambiguous interpretations. This can be overcome by giving a more specific description to an image, thereby preventing misinterpretations and providing more specific results. The proposed system aims at optimizing the search results by depicting more clearly the relationships between the objects in an image with the help of image captioning.