Geometry deep network image-set recognition method based on Grassmann manifolds

Tianci, 刘天赐 Liu; Zelin, 史泽林 Shi; Yunpeng, 刘云鹏 Liu; Yingdi, 张英迪 Zhang

doi:10.3788/irla201847.0703002

Cited by 2 publications

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…GrNet2B is composed of two Projection blocks, two Pooling blocks and one Output block. According to the research [50], the modified GrNet without Pooling Block can achieve higher accuracy, but it's significant to maintain the complete structure to reveal Grassmann manifold and helpful for feature work based on vanilla GrNet. The sizes of GrNet-2B weights are set to 400 × 300 and 150 × 100 to fit with the datasets and our method.…”

Section: Experiments Setupmentioning

confidence: 99%

Knowledge Distillation of Grassmann Manifold Network for Remote Sensing Scene Classification

Tian

Wang

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Due to device limitations, small networks are necessary for some real-world scenarios, such as satellites and micro-robots. Therefore, the development of a network with both good performance and small size is an important area of research. Deep networks can learn well from large amounts of data, while manifold networks have outstanding feature representation at small sizes. In this paper, we propose an approach that exploits the advantages of deep networks and shallow Grassmannian manifold networks. Inspired by knowledge distillation, we use the information learned from convolutional neural networks to guide the training of the manifold networks. Our approach leads to a reduction in model size, which addresses the problem of deploying deep learning on resource-limited embedded devices. Finally, a series of experiments were conducted on four remote sensing scene classification datasets. The method in this paper improved the classification accuracy by 2.31% and 1.73% on the UC Merced Land Use and SIRIWHU datasets, respectively, and the experimental results demonstrate the effectiveness of our approach.

show abstract

Section: Experiments Setupmentioning

confidence: 99%