.png){kind=icon}
返回
RTLinearFormer: Semantic segmentation with lightweight linear attentions
.png){kind=icon}
PRE
AI
DOI：10.1016/j.neucom.2025.129489.png)
.png){kind=icon}
.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Real-time semantic segmentation aims to balance networks' segmentation accuracy and inference speed. Recently, dual-resolution network shows its efficiency and effectiveness in this field. However, due to selfattention's O(N2) computational complexity, pure CNN-based networks still dominate in this architecture. In this paper, we propose RTLinearFormer, an efficient dual-resolution network with Transformer specially for real-time semantic segmentation, which achieves trade-off between segmentation accuracy and inference speed. To this end, our RTLinearFormer leverages lightweight channel-wise multi-scale attention with linear complexity, channel feature focusing and multi-scale features to extract global contextual information in low- resolution branch and uses cross-resolution ReLU-based linear attention with linear complexity to gather global contextual information derived from low-resolution branch with detailed information in high-resolution branch. Extensive experiments demonstrate the effectiveness and efficiency of our proposed RTLinearFormer in realtime semantic segmentation task. The model achieves 78.41% val mIoU and 66.7 FPS on Cityscapes using a single RTX 3090 GPU. When pretrained on Cityscapes, it achieves 77.4% test mIoU and 141.3 FPS on CamVid. It shows promising results on Cityscapes and CamVid, demonstrating superior performance over other stateof-the-art models. To further validate its generalization ability, we also test it on the ADE20K and COCOStuff datasets. It achieves 35.70% val mIoU and 107.3 FPS on ADE20K, and 31.15% test mIoU and 150.2 FPS on COCOStuff, setting new benchmarks for state-of-the-art performance.
Keyword:
Dual-resolution network
Lightweight channel-wise multi-scale attention
Cross-resolution ReLU-based linear attention
Real-time semantic segmentation
期刊
IF:
6.5
论文数:
2.5W
被引数:
6.5W
机构
引用论文
Caw’s Walking State Recognition Based on Accelerometers and Gyroscopes Installed on Ear-Tags and Collar-Tags基于安装在耳标和项圈上的加速度计和陀螺仪的Caw步行状态识别
.png){kind=icon}
PREAI
分享收藏Mist Control at a Machining Center, Part 2: Mist Control Following Installation of Air Cleaners
PREAI
分享收藏Inflexibility of mental planning: A characteristic disorder with prefrontal lobe lesions?心理计划的僵化: 前额叶病变的特征性障碍？
PREAI
分享收藏Distant regulatory elements in a Sox10‐βGEO BAC transgene are required for expression of Sox10 in the enteric nervous system and other neural crest‐derived tissuesSox10-βgeo BAC转基因中的远距离调控元件是肠神经系统和其他神经源性组织中 Sox10 表达所必需的
OAAI
OAAI分享收藏CO2 to Terpenes: Autotrophic and Electroautotrophic α‐Humulene Production with Cupriavidus necatorCO2 到萜烯: 使用 Cupriavidus necator 的自养和电自养 α-腐殖质生产
PREAI
分享收藏
查看更多内容