Contents Based Video Retrieval, and Summarization Using Deep Learning

Abstract

While digitization has changed the workflow of professional multimedia production, the content-based labeling of image sequences and video footage, necessary for all consequent stages of multimedia information, archival or marketing is typically still performed manually and thus quite time-consuming. This paper deals a broad analysis on the accessibility of video representations based on convolutional neural network (ConvNets) for the task of content base video retrieval. However, content based video retrieval prone to error, complex and time consuming with low accuracy using different traditional techniques. Other than the decision of convolutional layers, we display a proficient pipeline misusing multi-scale plans to extract hand crafted features, for selection of key-frames through clustering techniques. In the next step extract high level features from selected key-frames and evaluate the proposed search engine system for preliminary experiments. In our research, utilizing three standard videos datasets, our reveal  ConvNet for video representations can beat other state of the art, techniques that they are separated properly. This paper focuses on visual contents based video retrieval using deep features assessments generally uses of digital image processing, and the research improvement takes accounted for a view of a several databases.