Generating a textual description for an image is called image captioning. It is performed in two steps, Computer Vision and Natural Language Processing. Computer vision is used to detect the image and understand the context of the image. Natural Language Processing to turn the image into words called captions with the correct sequence. The automatic generation of captions for images is a challenging problem for researches. Content Based Image Retrieval (CBIR) which is a semantic based image retrieval matches user query based on some perceptual contents rather than text based query. The main issue in CBIR is to reduce the semantic gap. The role of image captioning in content based image retrieval is to achieve semantic similarity by capturing the textual data associated with the images. This paper focuses on image caption based processing system using deep learning for CBIR application. Convolutional neural network is used for image classification and Recurrent Neural network is used for caption generation. The resulting caption of the Image is used for retrieving similar images.
Volume 11 | 03-Special Issue
Pages: 1326-1331