建筑
计算机科学
变压器
编码器
网络体系结构
卷积神经网络
人工智能
编码(集合论)
计算机体系结构
实时计算
模式识别(心理学)
计算机网络
工程类
电气工程
程序设计语言
操作系统
地理
考古
电压
集合(抽象数据类型)
作者
Wele Gedara Chaminda Bandara,Vishal M. Patel
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2022-07-17
卷期号:: 207-210
被引量:1536
标识
DOI:10.1109/igarss46834.2022.9883686
摘要
This paper presents a transformer-based Siamese network architecture (abbreviated by ChangeFormer) for Change Detection (CD) from a pair of co-registered remote sensing images. Different from recent CD frameworks, which are based on fully convolutional networks (ConvNets), the proposed method unifies hierarchically structured transformer encoder with Multi-Layer Perception (MLP) decoder in a Siamese network architecture to efficiently render multi-scale long-range details required for accurate CD. Experiments on two CD datasets show that the proposed end-to-end trainable ChangeFormer architecture achieves better CD performance than previous counterparts. Our code and pre-trained models are available at github.com/wgcban/ChangeFormer.
科研通智能强力驱动
Strongly Powered by AbleSci AI