Automatic Image Caption Generation

Abstract

Recent work in machine learning uses an attention-based model, which automatically learns to predict the content of image. Automated Image Caption generation uses a type of artificial intelligence called deep learning to generate appropriate caption for the images. It also provides the best words to explain the image entirely. The textual description of the image would be generated after processing the image and its visual content has been analyzed. Caption generation can talk about the features of the scene and how the people and the objects in the image interact. The shape information of an image is mostly enclosed in edges. So, it is necessary to detect edges for an input image, by using certain filters and by enhancing those areas of image which contains edges, sharpness of image will increase and image will become clearer. The filter used here is sobel operator. Edge detection is used to extract features from the image and based on these features, caption will be generated which depicts the image. To convert these features into a meaningful caption, KNN classifier is used.