CNN-VWII: An Efficient Approach for Large-Scale Video Retrieval by Image Queries

TL;DR

This paper introduces CNN-VWII, a novel large-scale video retrieval method combining CNN and BoVW, utilizing a visual weighted inverted index to significantly improve retrieval speed and accuracy for image-to-video queries.

Contribution

It proposes a new model integrating CNN and BoVW for video frame representation and a visual weighted inverted index to enhance retrieval efficiency and accuracy.

Findings

Achieves up to tenfold speed improvements over existing methods.

Maintains comparable accuracy with state-of-the-art techniques.

Demonstrates effectiveness on large-scale video datasets.

Abstract

This paper aims to solve the problem of large-scale video retrieval by a query image. Firstly, we define the problem of top-$k$ image to video query. Then, we combine the merits of convolutional neural networks(CNN for short) and Bag of Visual Word(BoVW for short) module to design a model for video frames information extraction and representation. In order to meet the requirements of large-scale video retrieval, we proposed a visual weighted inverted index(VWII for short) and related algorithm to improve the efficiency and accuracy of retrieval process. Comprehensive experiments show that our proposed technique achieves substantial improvements (up to an order of magnitude speed up) over the state-of-the-art techniques with similar accuracy.