Domain Adaptation for Convolutional Neural Networks Based Remote Sensing Scene Classification

Journal Article · Mon Feb 25 00:00:00 EST 2019 · IEEE Geoscience and Remote Sensing Letters

DOI:https://doi.org/10.1109/LGRS.2019.2896411· OSTI ID:1492766

Song, Shaoyue ^[1]; Yu, Hongkai ^[2]; Miao, Zhenjiang ^[1]; Zhang, Qiang ^[1]; Lin, Yuewei ^[3]; Wang, Song ^[4]

---

1. Beijing Jiaotong Univ., Beijing (China)
2. Univ. of Texas, Edinburg, TX (United States)
3. Brookhaven National Lab. (BNL), Upton, NY (United States)
4. Univ. of South Carolina, Columbia, SC (United States); Tianjin Univ. (China)

Remote sensing (RS) scene classification plays an important role in the field of earth observation. With the rapid development of the remote sensing techniques, a large number of RS scene images are available. Since manually labeling largescale RS scene images is both labor and time consuming, when a new unlabeled dataset is obtained, how to use the existing labeled datasets to classify the new unlabeled images is an important research direction. Different RS scene image datasets may be taken from different type of sensors, and the images may vary from imaging modalities, spatial resolutions and image scales, so the distribution discrepancy exists among different image datasets. As a result, simply applying Convolutional Neural Networks (CNN) trained on source domain cannot accurately classify the images on target domain. Domain Adaptation (DA) can be helpful to solve this problem. In this paper, we design a Subspace Alignment (SA) and CNN based framework to solve the DA problem in RS scene image classification. A new SA layer is proposed and added into CNN models for DA, which could align the source and target domains in some feature subspace. Finetuning the modified CNN model with the added SA layer is able to make the CNN model adapt to the aligned feature subspace and helps to relieve the domain distribution discrepancy. The experiments conducted on two public datasets show that adding the SA layer into CNN can improve the scene classification on target domain.

• Research Organization: Brookhaven National Laboratory (BNL), Upton, NY (United States)
• Sponsoring Organization: USDOE Office of Science (SC), Advanced Scientific Computing Research (SC-21)
• Grant/Contract Number: SC0012704
• OSTI ID: 1492766
• Report Number(s): BNL--210923-2019-JAAM
• Journal Information: IEEE Geoscience and Remote Sensing Letters, Journal Name: IEEE Geoscience and Remote Sensing Letters Journal Issue: 8 Vol. 16; ISSN 1545-598X
• Publisher: IEEE
• Country of Publication: United States
• Language: English

Related Subjects

97 MATHEMATICS AND COMPUTING