Abstract | ||
---|---|---|
Synthetic datasets play a critical role in pre-training CNN models for optical flow, but they are painstaking to generate and hard to adapt to new applications. To automate the process, we present AutoFlow, a simple and effective method to render training data for optical flow that optimizes the performance of a model on a target dataset. AutoFlow takes a layered approach to render synthetic data, where the motion, shape, and appearance of each layer are controlled by learnable hyperparameters. Experimental results show that AutoFlow achieves state-of-the-art accuracy in pre-training both PWC-Net and RAFT. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1109/CVPR46437.2021.00996 | 2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021 |
DocType | ISSN | Citations |
Conference | 1063-6919 | 1 |
PageRank | References | Authors |
0.35 | 0 | 9 |
Name | Order | Citations | PageRank |
---|---|---|---|
Deqing Sun | 1 | 1061 | 44.84 |
Daniel Vlasic | 2 | 1 | 1.03 |
Charles Herrmann | 3 | 12 | 2.25 |
Varun Jampani | 4 | 184 | 19.44 |
Michael Krainin | 5 | 1 | 0.69 |
Huiwen Chang | 6 | 26 | 4.73 |
Ramin Zabih | 7 | 12976 | 982.19 |
William T. Freeman | 8 | 17382 | 1968.76 |
Ce Liu | 9 | 3347 | 188.04 |