Abstract | ||
---|---|---|
Fine-grained categorization refers to the task of categorizing objects with similar pattern or visual appearances. This study proposes a framework for feature evaluation and optimized selection. A random forest approach is employed to determine the importance of features and perform the dimension reduction by principle component analysis (PCA). The first four significant features representing shape, texture and color, respectively. The proposed framework can analyze the effectiveness of the features and achieve the balance between accuracy and computing time. The performance was evaluated on both ICL dataset and the self-constructed dataset. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1145/3055635.3056651 | ICMLC |
Field | DocType | ISBN |
Categorization,Dimensionality reduction,Pattern recognition,Feature selection,Computer science,Feature (computer vision),Feature evaluation,Feature extraction,Artificial intelligence,Random forest,Machine learning,Principal component analysis | Conference | 978-1-4503-4817-1 |
Citations | PageRank | References |
0 | 0.34 | 13 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Chia-Hung Wei | 1 | 0 | 0.68 |
Dapeng Zhang | 2 | 3 | 3.41 |
Yue Li | 3 | 0 | 0.34 |
Wei Wang | 4 | 1474 | 152.25 |