Transferable network with Siamese architecture for anomaly detection in hyperspectral images
International Journal of Applied Earth Observation and Geoinformation(2022)
摘要
•The anomaly detection problem is transformed into the similarity metric learning problem.•An adaptive unsupervised clustering process is proposed to generate pseudo labels from unlabeled images.•Pre-training and Fine-tuning strategies are adopted to ensure the transfer capability of the network.
更多查看译文
关键词
Hyperspectral images,Anomaly detection,Siamese architecture,Spectral-angle-based contrastive loss,Fine-tuning
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要