Bayesian Exponential Inverse Document Frequency and Region-of-Interest Effect for Enhancing Instance Search Accuracy

Masaya MURATA; Hidehisa NAGANO; Kaoru HIRAMATSU; Kunio KASHINO; Shin'ichi SATOH

doi:10.1587/transinf.2016EDP7066

Bayesian Exponential Inverse Document Frequency and Region-of-Interest Effect for Enhancing Instance Search Accuracy

Masaya MURATA, Hidehisa NAGANO, Kaoru HIRAMATSU, Kunio KASHINO, Shin'ichi SATOH

Full Text Views

0

Cite this

Summary :

In this paper, we first analyze the discriminative power in the Best Match (BM) 25 formula and provide its calculation method from the Bayesian point of view. The resulting, derived discriminative power is quite similar to the exponential inverse document frequency (EIDF) that we have previously proposed [1] but retains more preferable theoretical advantages. In our previous paper [1], we proposed the EIDF in the framework of the probabilistic information retrieval (IR) method BM25 to address the instance search task, which is a specific object search for videos using an image query. Although the effectiveness of our EIDF was experimentally demonstrated, we did not consider its theoretical justification and interpretation. We also did not describe the use of region-of-interest (ROI) information, which is supposed to be input to the instance search system together with the original image query showing the instance. Therefore, here, we justify the EIDF by calculating the discriminative power in the BM25 from the Bayesian viewpoint. We also investigate the effect of the ROI information for improving the instance search accuracy and propose two search methods incorporating the ROI effect into the BM25 video ranking function. We validated the proposed methods through a series of experiments using the TREC Video Retrieval Evaluation instance search task dataset.

Publication: IEICE TRANSACTIONS on Information Vol.E99-D No.9 pp.2320-2331

Publication Date: 2016/09/01

Publicized: 2016/06/03

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2016EDP7066

Type of Manuscript: PAPER

Category: Image Processing and Video Processing

Authors

Masaya MURATA
  NTT Corporation
Hidehisa NAGANO
  NTT Corporation
Kaoru HIRAMATSU
  NTT Corporation
Kunio KASHINO
  NTT Corporation,National Institute of Informatics
Shin'ichi SATOH
  National Institute of Informatics

Keyword

video retrieval, instance search, region-of-interest, TRECVID, probabilistic information retrieval, BM25, exponential IDF, Bayesian exponential IDF

Cite this

Copy

Masaya MURATA, Hidehisa NAGANO, Kaoru HIRAMATSU, Kunio KASHINO, Shin'ichi SATOH, "Bayesian Exponential Inverse Document Frequency and Region-of-Interest Effect for Enhancing Instance Search Accuracy" in IEICE TRANSACTIONS on Information, vol. E99-D, no. 9, pp. 2320-2331, September 2016, doi: 10.1587/transinf.2016EDP7066.
Abstract: In this paper, we first analyze the discriminative power in the Best Match (BM) 25 formula and provide its calculation method from the Bayesian point of view. The resulting, derived discriminative power is quite similar to the exponential inverse document frequency (EIDF) that we have previously proposed [1] but retains more preferable theoretical advantages. In our previous paper [1], we proposed the EIDF in the framework of the probabilistic information retrieval (IR) method BM25 to address the instance search task, which is a specific object search for videos using an image query. Although the effectiveness of our EIDF was experimentally demonstrated, we did not consider its theoretical justification and interpretation. We also did not describe the use of region-of-interest (ROI) information, which is supposed to be input to the instance search system together with the original image query showing the instance. Therefore, here, we justify the EIDF by calculating the discriminative power in the BM25 from the Bayesian viewpoint. We also investigate the effect of the ROI information for improving the instance search accuracy and propose two search methods incorporating the ROI effect into the BM25 video ranking function. We validated the proposed methods through a series of experiments using the TREC Video Retrieval Evaluation instance search task dataset.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2016EDP7066/_p

Copy

@ARTICLE{e99-d_9_2320,
author={Masaya MURATA, Hidehisa NAGANO, Kaoru HIRAMATSU, Kunio KASHINO, Shin'ichi SATOH, },
journal={IEICE TRANSACTIONS on Information},
title={Bayesian Exponential Inverse Document Frequency and Region-of-Interest Effect for Enhancing Instance Search Accuracy},
year={2016},
volume={E99-D},
number={9},
pages={2320-2331},
abstract={In this paper, we first analyze the discriminative power in the Best Match (BM) 25 formula and provide its calculation method from the Bayesian point of view. The resulting, derived discriminative power is quite similar to the exponential inverse document frequency (EIDF) that we have previously proposed [1] but retains more preferable theoretical advantages. In our previous paper [1], we proposed the EIDF in the framework of the probabilistic information retrieval (IR) method BM25 to address the instance search task, which is a specific object search for videos using an image query. Although the effectiveness of our EIDF was experimentally demonstrated, we did not consider its theoretical justification and interpretation. We also did not describe the use of region-of-interest (ROI) information, which is supposed to be input to the instance search system together with the original image query showing the instance. Therefore, here, we justify the EIDF by calculating the discriminative power in the BM25 from the Bayesian viewpoint. We also investigate the effect of the ROI information for improving the instance search accuracy and propose two search methods incorporating the ROI effect into the BM25 video ranking function. We validated the proposed methods through a series of experiments using the TREC Video Retrieval Evaluation instance search task dataset.},
keywords={},
doi={10.1587/transinf.2016EDP7066},
ISSN={1745-1361},
month={September},}

Copy

TY - JOUR
TI - Bayesian Exponential Inverse Document Frequency and Region-of-Interest Effect for Enhancing Instance Search Accuracy
T2 - IEICE TRANSACTIONS on Information
SP - 2320
EP - 2331
AU - Masaya MURATA
AU - Hidehisa NAGANO
AU - Kaoru HIRAMATSU
AU - Kunio KASHINO
AU - Shin'ichi SATOH
PY - 2016
DO - 10.1587/transinf.2016EDP7066
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E99-D
IS - 9
JA - IEICE TRANSACTIONS on Information
Y1 - September 2016
AB - In this paper, we first analyze the discriminative power in the Best Match (BM) 25 formula and provide its calculation method from the Bayesian point of view. The resulting, derived discriminative power is quite similar to the exponential inverse document frequency (EIDF) that we have previously proposed [1] but retains more preferable theoretical advantages. In our previous paper [1], we proposed the EIDF in the framework of the probabilistic information retrieval (IR) method BM25 to address the instance search task, which is a specific object search for videos using an image query. Although the effectiveness of our EIDF was experimentally demonstrated, we did not consider its theoretical justification and interpretation. We also did not describe the use of region-of-interest (ROI) information, which is supposed to be input to the instance search system together with the original image query showing the instance. Therefore, here, we justify the EIDF by calculating the discriminative power in the BM25 from the Bayesian viewpoint. We also investigate the effect of the ROI information for improving the instance search accuracy and propose two search methods incorporating the ROI effect into the BM25 video ranking function. We validated the proposed methods through a series of experiments using the TREC Video Retrieval Evaluation instance search task dataset.
ER -