Abstract | ||
---|---|---|
Recognizing figures in portrait Thangka images is fundamental to the appreciation of this unique Tibetan art. In this paper, we focus on the problem of portrait Thangka image retrieval from the perspective of re-identifying the figures in the images. Based on state-of-the-art re-identification methods, we further improve them by exploiting several tricks. We also investigate the impact of different cropping methods to evaluate the contribution of different features. Our evaluation results on an annotated portrait Thangka image dataset collected by ourselves demonstrate the necessity of further study on this challenging problem. We will release the data and code to promote the research on Thangka. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1007/978-3-030-86608-2_9 | BIOMETRIC RECOGNITION (CCBR 2021) |
Keywords | DocType | Volume |
Portrait Thangka, Image retrieval, Re-identification | Conference | 12878 |
ISSN | Citations | PageRank |
0302-9743 | 0 | 0.34 |
References | Authors | |
0 | 10 |
Name | Order | Citations | PageRank |
---|---|---|---|
Xire Danzeng | 1 | 0 | 0.34 |
Yuchao Yang | 2 | 0 | 1.35 |
Yufan Yang | 3 | 0 | 0.34 |
Zhao Hou | 4 | 0 | 0.34 |
Rui Xi | 5 | 0 | 0.34 |
Xinsheng Li | 6 | 0 | 0.34 |
Qijun Zhao | 7 | 419 | 38.37 |
Pubu Danzeng | 8 | 0 | 0.34 |
Gesang Duoji | 9 | 0 | 0.34 |
Dingguo Gao | 10 | 0 | 0.34 |