arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2202.09741cs.CV
收藏

视觉注意力网络

Visual Attention Network

Tsinghua University(清华大学);Nankai University(南开大学);Fitten Tech(Fitten科技)

Meng-Hao Guo, Cheng-Ze Lu, Zheng-Ning Liu, Ming-Ming Cheng, Shi-Min Hu

摘要:

虽然最初是为自然语言处理任务设计的,但自我注意力机制最近在各种计算机视觉领域掀起了风暴。然而,图像的2D性质为在计算机视觉中应用自注意力带来了三个挑战。(1)将图像视为1D序列忽略了它们的2D结构。(2)二次复杂度对于高分辨率图像来说太昂贵了。(3)它只考虑了空间适应性,而忽略了信道适应性。在本文中,我们提出了一种新型的线性注意力,称为大内核注意力(LKA),以实现自我注意力的自适应和远程相关性,同时避免其缺点。在此基础上,提出了一种基于LKA的神经网络,即视觉注意力网络(VAN)。虽然非常简单,但VAN在各种任务中超越了类似尺寸的视觉Transformer(ViT)和卷积神经网络(CNN),包括图像分类,目标检测,语义分割,全景分割,姿态估计等。例如,VAN-B6在ImageNet基准测试中达到了87.8%的准确率,并为全景分割设置了新的最先进的性能(58.2 PQ)。此外,VAN-B2在ADE 20 K基准测试中的语义分割方面超过Swin-T 4% mIoU(50.1 vs. 46.1),在COCO数据集上的目标检测方面超过Swin-T 2.6% AP(48.8 vs. 46.2)。它为社区提供了一种新颖的方法和一个简单而强大的基线。代码可在https://github.com/Visual-Attention-Network上获得。

英文摘要:

While originally designed for natural language processing tasks, the self-attention mechanism has recently taken various computer vision areas by storm. However, the 2D nature of images brings three challenges for applying self-attention in computer vision. (1) Treating images as 1D sequences neglects their 2D structures. (2) The quadratic complexity is too expensive for high-resolution images. (3) It only captures spatial adaptability but ignores channel adaptability. In this paper, we propose a novel linear attention named large kernel attention (LKA) to enable self-adaptive and long-range correlations in self-attention while avoiding its shortcomings. Furthermore, we present a neural network based on LKA, namely Visual Attention Network (VAN). While extremely simple, VAN surpasses similar size vision transformers(ViTs) and convolutional neural networks(CNNs) in various tasks, including image classification, object detection, semantic segmentation, panoptic segmentation, pose estimation, etc. For example, VAN-B6 achieves 87.8% accuracy on ImageNet benchmark and set new state-of-the-art performance (58.2 PQ) for panoptic segmentation. Besides, VAN-B2 surpasses Swin-T 4% mIoU (50.1 vs. 46.1) for semantic segmentation on ADE20K benchmark, 2.6% AP (48.8 vs. 46.2) for object detection on COCO dataset. It provides a novel method and a simple yet strong baseline for the community. Code is available at https://github.com/Visual-Attention-Network.

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。

↑