Name
Title | ||
|---|---|---|
Cross-Database Micro-Expression Recognition: A Style Aggregated and Attention Transfer Approach |
Abstract | ||
|---|---|---|
Deep learning systems, such as the Residual Network (ResNet), can infer a hierarchical representation of input data that facilitates categorization. In this paper, we propose a Style Aggregated and Attention Transfer framework (SA-AT) based on ResNet for cross-database Micro-Expression Recognition (MER). The training of SA-AT has two stages. In the first stage, facial expression samples are used as the auxiliary database to train a ResNet teacher model. To benefit from the significant classification ability of teacher model which is trained on large scale data, in the second step, the attention of the teacher model is transferred to train the student model on style aggregated micro-expression databases with limited samples. Our experiments demonstrate that compared with the baseline method in Micro-Expression Grand Challenge 2019, our proposed technique achieves more promising performance. |
Year | DOI | Venue |
|---|---|---|
2019 | 10.1109/ICMEW.2019.00025 | 2019 IEEE International Conference on Multimedia & Expo Workshops (ICMEW) |
Keywords | Field | DocType |
cross-database micro-expression recognition, micro-expression recognition, transfer learning | Categorization,Residual,Pattern recognition,Facial expression recognition,Computer science,Transfer of learning,Facial expression,Artificial intelligence,Deep learning,Residual neural network,Database | Conference |
ISSN | ISBN | Citations |
2330-7927 | 978-1-5386-9215-8 | 1 |
PageRank | References | Authors |
0.35 | 6 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Ling Zhou | 1 | 1 | 0.68 |
| Qirong Mao | 2 | 261 | 34.29 |
| Luoyang Xue | 3 | 2 | 1.36 |