计算机应用与软件2018，Vol.35Issue(1)：160-163,182,5.DOI:10.3969/j.issn.1000-386x.2018.01.028
基于深度学习的视频近似拷贝检索
VIDEO NEAR-DUPLICATE RETRIEVAL BASED ON DEEP LEARNING
摘要
Abstract
With the development of internet and multimedia technology,there are more and more multimedia data on the Internet,detecting and removing duplicate content in database has become an important task.Based on this situation,this article designed a fast and effective video near-duplicate retrieval system.Unlike traditional approaches,our method adopts deep Convolutional Neural Network to extract frame features,then we use k-d tree to build index for the features,which rapidly speeds up the following searching step.Finally we calculate the similarities between the query video and each candidate and consider the sorted similarities as the final result.关键词
近似拷贝检索/特征值提取/深度学习/k-d树/相似度度量Key words
Near-duplicate retrieval/Feature extraction/Deep learning/K-d tree/Similarity measure分类
信息技术与安全科学引用本文复制引用
冯兆华,朱允斌,李卫强..基于深度学习的视频近似拷贝检索[J].计算机应用与软件,2018,35(1):160-163,182,5.基金项目
国家重点研发计划项目(2016YFC0801003) （2016YFC0801003）
上海市科委科研计划项目(16511105402) （16511105402）
上海市人才计划项目(17XD1425000). （17XD1425000）
