arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2407.17261cs.CV
收藏

具有推理空间缩减的高效语义分割无嵌入Transformer

Embedding-Free Transformer with Inference Spatial Reduction for Efficient Semantic Segmentation

Sogang University(西江大学);LG Electronics(LG电子);Pusan National University(釜山大学)

Hyunwoo Yu, Yubin Cho, Beoungwoo Kang, Seunghun Moon, Kyeongbo Kong, Suk-Ju Kang

AI总结:

本文提出EDAFormer,包含无嵌入Transformer编码器、全注意力解码器及推理空间缩减(ISR)方法,通过优化全局建模与降低推理阶段key-value分辨率,在多个基准上实现高效且先进的语义分割性能。

摘要:

我们提出了一种编码器-解码器注意力Transformer(EDAFormer),它由无嵌入Transformer(EFT)编码器和使用我们提出的无嵌入注意力(EFA)结构的全注意力解码器组成。所提出的EFA是一种新颖的全局上下文建模机制,其重点在于发挥全局非线性作用,而非query、key和value的具体角色。对于解码器,我们探索了考虑全局性的优化结构,这可以提高语义分割性能。此外,我们提出了一种新颖的推理空间缩减(ISR)方法以提升计算效率。与以往的空间缩减注意力方法不同,我们的ISR方法在推理阶段进一步降低了key-value的分辨率,这可以缓解高效语义分割中计算与性能之间的权衡差距。与现有的基于Transformer的语义分割模型相比,我们的EDAFormer在ADE20K、Cityscapes和COCO-Stuff三个公开基准上展现出具有高效计算的先进性能。此外,我们的ISR方法在Cityscapes数据集上将计算成本降低了高达61%,且仅有极小的mIoU性能下降。代码可在https://github.com/hyunwoo137/EDAFormer获取。

英文摘要:

We present an Encoder-Decoder Attention Transformer, EDAFormer, which consists of the Embedding-Free Transformer (EFT) encoder and the all-attention decoder leveraging our Embedding-Free Attention (EFA) structure. The proposed EFA is a novel global context modeling mechanism that focuses on functioning the global non-linearity, not the specific roles of the query, key and value. For the decoder, we explore the optimized structure for considering the globality, which can improve the semantic segmentation performance. In addition, we propose a novel Inference Spatial Reduction (ISR) method for the computational efficiency. Different from the previous spatial reduction attention methods, our ISR method further reduces the key-value resolution at the inference phase, which can mitigate the computation-performance trade-off gap for the efficient semantic segmentation. Our EDAFormer shows the state-of-the-art performance with the efficient computation compared to the existing transformer-based semantic segmentation models in three public benchmarks, including ADE20K, Cityscapes and COCO-Stuff. Furthermore, our ISR method reduces the computational cost by up to 61% with minimal mIoU performance degradation on Cityscapes dataset. The code is available at https://github.com/hyunwoo137/EDAFormer.

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。

↑