Automatic determination of image features for material perception by machine learning and construction of the perception space
Project/Area Number |
26730084
|
Research Category |
Grant-in-Aid for Young Scientists (B)
|
Allocation Type | Multi-year Fund |
Research Field |
Perceptual information processing
|
Research Institution | Chiba University |
Principal Investigator |
Yata Noriko 千葉大学, 大学院融合科学研究科, 助教 (60528412)
|
Project Period (FY) |
2014-04-01 – 2017-03-31
|
Project Status |
Completed (Fiscal Year 2016)
|
Budget Amount *help |
¥3,770,000 (Direct Cost: ¥2,900,000、Indirect Cost: ¥870,000)
Fiscal Year 2016: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2015: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2014: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
|
Keywords | 質感 / 視覚情報処理 / ニューラルネットワーク / 機械学習 / 画像 / 特徴量 / 進化計算 / 関数同定 / 金属感 / 主観評価実験 / 評定尺度法 / 主成分分析 / 回帰分析 |
Outline of Final Research Achievements |
This research focuses on the metallic perception that people feel. First, CG images ware created based on the reflection characteristics. Next, values of metal perception of each image ware measured using a subjective evaluation experiment. The results of this experiment ware analyzed for deciding the axis of the metallic texture scale. The constructed metallic texture scale was evaluated by examining whether the numerical value of the parameter correspond to the metallic appearance. We have also proposed a method to analyze recognition mechanism of CNNs. We obtained material recognition CNNs that have sufficiently meaningful for analysis. By visualizing the hidden layer of CNNs, we found that a CNN’s recognition process differs depending on the presence or absence of texture.
|
Report
(4 results)
Research Products
(12 results)