论文标题
筏：复发全对场的光流转换
RAFT: Recurrent All-Pairs Field Transforms for Optical Flow
论文作者
论文摘要
我们介绍了反复的全对场变换（RAFT），这是一种用于光流的新的深网架构。 RAFT提取物每像素功能，为所有像素对构建多尺度的4D相关量，并迭代地通过在相关量上执行查找的复发单元更新流场。木筏实现最先进的性能。在KITTI上，木筏的F1误差为5.10％，与最佳发布结果（6.10％）相比，误差降低了16％。在Sintel（最终通过）上，RAFT获得了2.855像素的终点误差，从最佳发布的结果（4.098像素）降低了30％的误差。此外，木筏在推理时间，训练速度和参数计数方面具有强大的跨数据集泛化以及高效率。代码可在https://github.com/princeton-vl/raft上找到。
We introduce Recurrent All-Pairs Field Transforms (RAFT), a new deep network architecture for optical flow. RAFT extracts per-pixel features, builds multi-scale 4D correlation volumes for all pairs of pixels, and iteratively updates a flow field through a recurrent unit that performs lookups on the correlation volumes. RAFT achieves state-of-the-art performance. On KITTI, RAFT achieves an F1-all error of 5.10%, a 16% error reduction from the best published result (6.10%). On Sintel (final pass), RAFT obtains an end-point-error of 2.855 pixels, a 30% error reduction from the best published result (4.098 pixels). In addition, RAFT has strong cross-dataset generalization as well as high efficiency in inference time, training speed, and parameter count. Code is available at https://github.com/princeton-vl/RAFT.
