Pub Date : 2010-09-21DOI: 10.1109/CBMI.2011.5972514
R. Tavenard, H. Jégou, L. Amsaleg
Many algorithms for approximate nearest neighbor search in high-dimensional spaces partition the data into clusters. At query time, for efficiency, an index selects the few (or a single) clusters nearest to the query point. Clusters are often produced by the well-known k-means approach since it has several desirable properties. On the downside, it tends to produce clusters having quite different cardinalities. Imbalanced clusters negatively impact both the variance and the expectation of query response times. This paper proposes to modify k-means centroids to produce clusters with more comparable sizes without sacrificing the desirable properties. Experiments with a large scale collection of image descriptors show that our algorithm significantly reduces the variance of response times without severely impacting the search quality.
{"title":"Balancing clusters to reduce response time variability in large scale image search","authors":"R. Tavenard, H. Jégou, L. Amsaleg","doi":"10.1109/CBMI.2011.5972514","DOIUrl":"https://doi.org/10.1109/CBMI.2011.5972514","url":null,"abstract":"Many algorithms for approximate nearest neighbor search in high-dimensional spaces partition the data into clusters. At query time, for efficiency, an index selects the few (or a single) clusters nearest to the query point. Clusters are often produced by the well-known k-means approach since it has several desirable properties. On the downside, it tends to produce clusters having quite different cardinalities. Imbalanced clusters negatively impact both the variance and the expectation of query response times. This paper proposes to modify k-means centroids to produce clusters with more comparable sizes without sacrificing the desirable properties. Experiments with a large scale collection of image descriptors show that our algorithm significantly reduces the variance of response times without severely impacting the search quality.","PeriodicalId":358337,"journal":{"name":"2011 9th International Workshop on Content-Based Multimedia Indexing (CBMI)","volume":"51 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2010-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121444585","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 1900-01-01DOI: 10.1109/CBMI.2011.5972539
Yingbo Li, B. Mérialdo
In this paper we propose a novel algorithm for video summarization, OB-MMR (Optimized Balanced Audio Video Maximal Marginal Relevance). This algorithm is suitable to summarize both single and multiple videos. OB-MMR is achieved by optimizing the parameters in Balanced AV-MMR (Balanced Audio Video Maximal Marginal Relevance), namely the balance factor between audio information and visual information in the video, but also the importance of face and audio transitions among audio segments with different genres. Therefore, OB-MMR achieves a better result than previous algorithms, Video-MMR and Balanced AV-MMR. Furthermore, it is possible to select the optimized parameters for each genre of videos, which leads to promising automatic algorithms for video summarization in the future large-scale experiments.
在本文中，我们提出了一种新的视频摘要算法，OB-MMR(优化平衡音视频最大边际相关性)。该算法既适用于单个视频，也适用于多个视频的总结。OB-MMR是通过优化Balanced AV-MMR (Balanced Audio Video maximum Marginal Relevance，平衡音频视频最大边际相关性)中的参数来实现的，即视频中音频信息和视觉信息之间的平衡因子，以及不同类型音频片段之间人脸和音频过渡的重要性。因此，OB-MMR比之前的Video-MMR和Balanced AV-MMR算法取得了更好的效果。此外，它还可以为每个视频类型选择优化的参数，从而在未来的大规模实验中为视频摘要提供有前途的自动算法。
{"title":"Multi-video summarization based on OB-MMR","authors":"Yingbo Li, B. Mérialdo","doi":"10.1109/CBMI.2011.5972539","DOIUrl":"https://doi.org/10.1109/CBMI.2011.5972539","url":null,"abstract":"In this paper we propose a novel algorithm for video summarization, OB-MMR (Optimized Balanced Audio Video Maximal Marginal Relevance). This algorithm is suitable to summarize both single and multiple videos. OB-MMR is achieved by optimizing the parameters in Balanced AV-MMR (Balanced Audio Video Maximal Marginal Relevance), namely the balance factor between audio information and visual information in the video, but also the importance of face and audio transitions among audio segments with different genres. Therefore, OB-MMR achieves a better result than previous algorithms, Video-MMR and Balanced AV-MMR. Furthermore, it is possible to select the optimized parameters for each genre of videos, which leads to promising automatic algorithms for video summarization in the future large-scale experiments.","PeriodicalId":358337,"journal":{"name":"2011 9th International Workshop on Content-Based Multimedia Indexing (CBMI)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1900-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128718215","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：