| Project/Area Number |
15K16028
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Perceptual information processing
|
|---|
| Research Institution | Kyushu University |
|---|
Principal Investigator |
Matsukawa Tetsu 九州大学, システム情報科学研究院, 助教 (80747212)
|
|---|
| Project Period (FY) |
2015-04-01 – 2017-03-31
|
| Project Status |
Completed (Fiscal Year 2016)
|
| Budget Amount *help |
¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000)
Fiscal Year 2016: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2015: ¥2,470,000 (Direct Cost: ¥1,900,000、Indirect Cost: ¥570,000)
|
|---|
| Keywords | カメラ間人物照合 / 人物属性 / 深層特徴 / CNN / 監視カメラ / 歩行者 |
|---|
| Outline of Final Research Achievements |
Aiming at to apply for person re-identification, we have developed deep features based on various human attributes (clothing, carrying object etc.). We conducted a learning of Convolutional Neural Network (CNN) using a pedestrian dataset which is annotated by human attributes. We extracted intermediate features from the learnt CNN and transferred them to person re-identification. To obtain more discriminative features in CNN, we developed a method to learn CNN using combination of human attributes. This method produces fine-grained labels for the CNN learning by focusing on the attribute combinations among different attribute groups, eg., upper body clothing, lower body clothing, and their colors. The quantitative evaluation on the benchmark datasets confirmed the effectiveness of the proposed method.
|
