DB-BlendMask: Decomposed Attention and Balanced BlendMask for Instance Segmentation of High-Resolution Remote Sensing Images

• Published in: IEEE transactions on geoscience and remote sensing Vol. 60; pp. 1 - 15
• Main Authors: Chen, Zhenqian, Shang, Yongheng, Python, Andre, Cai, Yuxiang, Yin, Jianwei
• Format: Journal Article
• Language: English
• Published: New York IEEE 2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Artificial neural networks; Benchmarks; BlendMask; class balance module (CBM); Complexity; decomposed attention blender module (DA-Blender module); Decomposition; Detection; Feature extraction; High resolution; high-resolution remote sensing images (HRRSIs); Image processing; Image resolution; Image segmentation; Instance segmentation; Marine vehicles; Modules; Neural networks; Object detection; Object recognition; Remote sensing; Resolution; Resource management; Semantics; Shape; size balance coefficient (SBC)
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2021.3138913

Instance segmentation is an important method for high-resolution remote sensing images (HRRSIs) analysis. Traditional instance segmentation algorithms are not suitable to analyze complex HRRSIs that exhibit: 1) various shapes and sizes of targets; 2) a large number of small targets; and 3) data with long tail distribution. Here we introduce DB-BlendMask, an efficient and accurate instance segmentation method that can accommodate complex HRRSIs. It is composed of size balance coefficient (SBC), class balance module (CBM), and decomposed attention blender module (DA-Blender module). SBC consists of a fair weight allocation strategy for positive samples in object detection. CBM combines classification obtained in object detection stage to guide the semantic feature extraction. Complementary to a traditional convolutional neural network (CNN) architecture, DA-Blender module has the ability to considerably compress space complexity of attention and merge attention with semantic feature to generate the instance mask. We compare the performance of DB-BlendMask with a benchmark Mask R-CNN on two typical datasets, iSAID, and ISPRS Postdam. We obtain an average detection precision of 39.2% on iSAID and 63.6% on ISPRS Postdam, which corresponds to an improvement of 2.5% and 2.7%, respectively, compared to the benchmark in a real-time scenario.