Content based Video Retrieval on Key Frames using AleNet-Deep Learning Model
Javed Iqbal Bangash, Institute of Computer Sciences and Information Technology (ICS/IT), The University of Agriculture Peshawar, Pakistan.
Abdullah Khan, Institute of Computer Sciences and Information Technology (ICS/IT), The University of Agriculture Peshawar, Pakistan.
Abdul Waheed Khan, Institute of Computer Sciences and Information Technology (ICS/IT), The University of Agriculture Peshawar, Pakistan.
Mehtab Ahmad, Institute of Computer Sciences and Information Technology (ICS/IT), The University of Agriculture Peshawar, Pakistan.
Corresponding Author:
Abdul Waheed Khan (waheed.khan@fecid.paf-iast.edu.pk)
Abstract:
A video retrieval system is tasked with finding the most appropriate video collection in response to a user query. The contents of a movie can be extracted using several feature extraction models. The demand for Content Based Video Retrieval (CBVR) has arisen as a result of the exponential expansion in digital video databases, as well as their widespread deployment in applications ranging in health, military, social media, and art. The actual content of the video (Colors, Textures, and Shapes) is examined in the CBVR system, and similar films from the database are retrieved. Due to the prevalence of complex visual features, locating similar videos from a large video collection is a difficult operation. Color, form, and texture are all visual aspects of digital video. A content-based video retrieval system based on Convolutional Neural Networks (CNNs) on key frames has been proposed in this study. The proposed method was evaluated using the benchmark UCF101dataset. As compared to the state-of-the-art strategies, the proposed method showed better outcomes using accuracy and loss.
Keywords:
CNN; K-Mean; CBVR; Color Histogram; Accuracy; Loss; Alex Net