Fine-grained video retrieval from large-scale video using query sentences containing unknown concepts
| Project/Area Number |
18K11362
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 61010:Perceptual information processing-related
|
|---|
| Research Institution | Meisei University |
|---|
Principal Investigator |
Ueki Kazuya 明星大学, 情報学部, 准教授 (80580638)
|
|---|
| Project Period (FY) |
2018-04-01 – 2022-03-31
|
| Project Status |
Completed (Fiscal Year 2021)
|
| Budget Amount *help |
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2020: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2019: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2018: ¥1,950,000 (Direct Cost: ¥1,500,000、Indirect Cost: ¥450,000)
|
|---|
| Keywords | 映像検索 / クエリ文 / TRECVID / 未知の概念 / 画像/言語の同時埋め込み |
|---|
| Outline of Final Research Achievements |
We worked on a technique for retrieving videos that match unknown query sentences containing new concepts, such as newly emerging trends and new methods of crime, from the wide variety of videos uploaded to the Internet every day. We investigated methods to apply the trained models to video frames by embedding images and language using a large number of images with captions. We evaluated our method on a large scale of videos in the international competitive video retrieval and evaluation benchmark (TRECVID) organized by the National Institute of Standards and Technology (NIST), and confirmed that our method could retrieve videos with high accuracy for a wide range of query sentences.
|
| Academic Significance and Societal Importance of the Research Achievements |
ラベル付きの画像・映像データの整備と，ディープラーニング技術の進展に伴い，ある特定の物体・シーン・動作等のキーワードに合致した画像や映像の検索が実現されつつある．しかしながら，ライフスタイルの変化により，複数かつ新しい概念を含んだクエリ文を用いた詳細映像検索の実現が期待されている．本研究成果により，幅広く複雑なクエリ文に対して高精度に映像を検索することが可能であることから，新しく生まれる未知の概念が含まれている場合においても，その説明文をクエリとして入力することで，必要となる映像を即座に検索できることが期待される．
|
Report
(5 results)
Research Products
(19 results)
