论文标题
流动式：光流的变压器体系结构
FlowFormer: A Transformer Architecture for Optical Flow
论文作者
论文摘要
我们介绍了光流变压器，被称为流动型，这是一种基于变压器的神经网络体系结构，用于学习光流。流动形式将图像对构建的4D成本量构成，将成本令牌编码为具有备用组变压器（AGT）层的成本记忆，并在新型潜在空间中编码，并通过反复的变压器解码器和动态位置成本查询来解码成本记忆。在SINTEL基准测试中，FlowFormer在干净和最终通过时达到了1.159和2.088平均终点率（AEPE），从最佳发布的结果（1.388和2.47）降低了16.5％和15.5％的误差。此外，流程度还达到了强大的概括性能。在不接受Sintel的培训的情况下，FlowFormer在Sintel训练套装的清洁通行证上实现了1.01 AEPE，表现优于最佳发布结果（1.29）（1.29），提高了21.7％。
We introduce optical Flow transFormer, dubbed as FlowFormer, a transformer-based neural network architecture for learning optical flow. FlowFormer tokenizes the 4D cost volume built from an image pair, encodes the cost tokens into a cost memory with alternate-group transformer (AGT) layers in a novel latent space, and decodes the cost memory via a recurrent transformer decoder with dynamic positional cost queries. On the Sintel benchmark, FlowFormer achieves 1.159 and 2.088 average end-point-error (AEPE) on the clean and final pass, a 16.5% and 15.5% error reduction from the best published result (1.388 and 2.47). Besides, FlowFormer also achieves strong generalization performance. Without being trained on Sintel, FlowFormer achieves 1.01 AEPE on the clean pass of Sintel training set, outperforming the best published result (1.29) by 21.7%.
