Extention of TSCDP for Video Classification
をテンプレートにして作成
[
トップ
] [
新規
|
一覧
|
単語検索
|
最終更新
|
ヘルプ
|
ログイン
]
開始行:
[[Available Research Topics]]
*Extension of TSCDP for Video Classification [#v99dd35f]
In this research, you will develop a scheme to classify movies with the category defined for a benchmark. In the video recognition processing field, we can see many algorithms to classify movies in several categories as a kind of sport, behavior and scenes. The goal of this research is to high recognition rate than previous works. You establish an algorithm to learn the relationship between the specification of movies and categories. Time-series continuous dynamic programming (TSCDP) will be used to extract features. Our current status is;
+ feature extraction using TSCDP
+ Learning categories from future extraction
The key point of this research is TSCDP. You will modify TSCDP algorithm to improve the result of classification.
*Research Outline [#qf2396fa]
-Understanding TSCDP
-Survey of current video recognition algorithms
-Modification of reference traces to improve the result
-Modification of TSCDP for scaling
-Evaluation of the new TSCDP video recognition framework
*Expected Output [#l4203a3d]
-How to use video recognition framework with TSCDP (Tutorial)
-Summary of current video recognition algorithms
-How to modify DP process for scaling detection
-Evaluation summary of your algorithm using a benchmark
*Reference [#b7b0edc1]
- 画処理研卒業生のスライド:&ref(m5171141.pdf);
- TSCDPとその関連技術:&ref(paper.pdf);
- 2015年パルテノン研究会原稿:&ref(parthenon_hayamizu.docx);
終了行:
[[Available Research Topics]]
*Extension of TSCDP for Video Classification [#v99dd35f]
In this research, you will develop a scheme to classify movies with the category defined for a benchmark. In the video recognition processing field, we can see many algorithms to classify movies in several categories as a kind of sport, behavior and scenes. The goal of this research is to high recognition rate than previous works. You establish an algorithm to learn the relationship between the specification of movies and categories. Time-series continuous dynamic programming (TSCDP) will be used to extract features. Our current status is;
+ feature extraction using TSCDP
+ Learning categories from future extraction
The key point of this research is TSCDP. You will modify TSCDP algorithm to improve the result of classification.
*Research Outline [#qf2396fa]
-Understanding TSCDP
-Survey of current video recognition algorithms
-Modification of reference traces to improve the result
-Modification of TSCDP for scaling
-Evaluation of the new TSCDP video recognition framework
*Expected Output [#l4203a3d]
-How to use video recognition framework with TSCDP (Tutorial)
-Summary of current video recognition algorithms
-How to modify DP process for scaling detection
-Evaluation summary of your algorithm using a benchmark
*Reference [#b7b0edc1]
- 画処理研卒業生のスライド:&ref(m5171141.pdf);
- TSCDPとその関連技術:&ref(paper.pdf);
- 2015年パルテノン研究会原稿:&ref(parthenon_hayamizu.docx);
ページ名: