arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2108.06152cs.CV
收藏

用于快速训练收敛的 Conditional DETR

Conditional DETR for Fast Training Convergence

University of Science and Technology of China(中国科学技术大学);Peking University(北京大学);Microsoft Research Asia(微软亚洲研究院)

Depu Meng, Xiaokang Chen, Zejia Fan, Gang Zeng, Houqiang Li, Yuhui Yuan, Lei Sun, Jingdong Wang

AI总结:

针对 DETR 训练收敛慢的问题,本文提出 conditional DETR,通过从解码器嵌入学习条件空间查询来增强多头交叉注意力、降低对内容嵌入的依赖,并显著加快多种主干网络下的收敛。

摘要:

近期提出的 DETR 方法将 transformer 编码器和解码器架构应用于目标检测,并取得了很有前景的性能。本文针对其训练收敛缓慢这一关键问题,提出一种用于快速训练 DETR 的条件交叉注意力机制。该方法的动机是:DETR 中的交叉注意力高度依赖内容嵌入来定位四个端点并预测边界框,这提高了对高质量内容嵌入的需求,从而增加了训练难度。我们的方法名为 conditional DETR,它从解码器嵌入中学习一个条件空间查询,用于解码器多头交叉注意力。其优势在于,通过条件空间查询,每个交叉注意力头都能够关注包含一个不同区域的带状范围,例如一个目标端点或目标框内部的一个区域。这缩小了用于定位不同区域以进行目标分类和边界框回归的空间范围,从而放松了对内容嵌入的依赖并简化了训练。实验结果表明,在主干网络 R50 和 R101 上,conditional DETR 的收敛速度提升 6.7 倍;在更强的主干网络 DC5-R50 和 DC5-R101 上,收敛速度提升 10 倍。代码可在 https://github.com/Atten4Vis/ConditionalDETR 获取。

英文摘要:

The recently-developed DETR approach applies the transformer encoder and decoder architecture to object detection and achieves promising performance. In this paper, we handle the critical issue, slow training convergence, and present a conditional cross-attention mechanism for fast DETR training. Our approach is motivated by that the cross-attention in DETR relies highly on the content embeddings for localizing the four extremities and predicting the box, which increases the need for high-quality content embeddings and thus the training difficulty. Our approach, named conditional DETR, learns a conditional spatial query from the decoder embedding for decoder multi-head cross-attention. The benefit is that through the conditional spatial query, each cross-attention head is able to attend to a band containing a distinct region, e.g., one object extremity or a region inside the object box. This narrows down the spatial range for localizing the distinct regions for object classification and box regression, thus relaxing the dependence on the content embeddings and easing the training. Empirical results show that conditional DETR converges 6.7x faster for the backbones R50 and R101 and 10x faster for stronger backbones DC5-R50 and DC5-R101. Code is available at https://github.com/Atten4Vis/ConditionalDETR.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑