摘要
Image landmark detection aims to automatically identify the locations of predefined fiducial points. Despite recent success in this field, higher-ordered structural modeling to capture implicit or explicit relationships among anatomical landmarks has not been adequately exploited. In this work, we present a new topology-adapting deep graph learning approach for accurate anatomical facial and medical (e.g., hand, pelvis) landmark detection. The proposed method constructs graph signals leveraging both local image features and global shape features. The adaptive graph topology naturally explores and lands on task-specific structures which are learned end-to-end with two Graph Convolutional Networks (GCNs). Extensive experiments are conducted on three public facial image datasets (WFLW, 300W, and COFW-68) as well as three real-world X-ray medical datasets (Cephalometric (public), Hand and Pelvis). Quantitative results comparing with the previous state-of-the-art approaches across all studied datasets indicating the superior performance in both robustness and accuracy. Qualitative visualizations of the learned graph topologies demonstrate a physically plausible connectivity laying behind the landmarks.
| 原文 | 英語 |
|---|---|
| 主出版品物標題 | Computer Vision – ECCV 2020 - 16th European Conference, 2020, Proceedings |
| 編輯 | Andrea Vedaldi, Horst Bischof, Thomas Brox, Jan-Michael Frahm |
| 發行者 | Springer Science and Business Media Deutschland GmbH |
| 頁面 | 266-283 |
| 頁數 | 18 |
| ISBN(列印) | 9783030585440 |
| DOIs | |
| 出版品狀態 | 已出版 - 2020 |
| 事件 | 16th European Conference on Computer Vision, ECCV 2020 - Glasgow, 英國 持續時間: 23 08 2020 → 28 08 2020 |
出版品系列
| 名字 | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| 卷 | 12354 LNCS |
| ISSN(列印) | 0302-9743 |
| ISSN(電子) | 1611-3349 |
Conference
| Conference | 16th European Conference on Computer Vision, ECCV 2020 |
|---|---|
| 國家/地區 | 英國 |
| 城市 | Glasgow |
| 期間 | 23/08/20 → 28/08/20 |
文獻附註
Publisher Copyright:© 2020, Springer Nature Switzerland AG.
指紋
深入研究「Structured Landmark Detection via Topology-Adapting Deep Graph Learning」主題。共同形成了獨特的指紋。引用此
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver