Name
Abstract | ||
|---|---|---|
This paper proposes a new and effective convolutional neural network model termed OFR-Net for optical flow refinement. The OFR-Net exploits the spatial correlation between images and optical flow fields. It adopts a pyramidal codec structure with residual connections, dense connections and skip connections within and between the encoder and decoder, to comprehensively fuse features of different scales, locally and globally. We also introduce a warp loss to restrict large displacement refinement errors. A series of experiments on the FlyingChairs and MPI Sintel datasets show that the OFR-Net can effectively refine the optical flow predicted by various methods. |
Year | DOI | Venue |
|---|---|---|
2020 | 10.1587/transfun.2020EAL2024 | IEICE TRANSACTIONS ON FUNDAMENTALS OF ELECTRONICS COMMUNICATIONS AND COMPUTER SCIENCES |
Keywords | DocType | Volume |
optical flow refinement, pyramid dense residual structure, residual learning, spatial correlation, warp loss | Journal | E103A |
Issue | ISSN | Citations |
11 | 0916-8508 | 0 |
PageRank | References | Authors |
0.34 | 0 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Liping Zhang | 1 | 0 | 0.34 |
| Lu ZQ | 2 | 47 | 9.45 |
| QM | 3 | 464 | 72.05 |