MAP-Net: Multi Attending Path Neural Network for Building Footprint Extraction from Remote Sensed Imagery

Abstract

Accurately and efficiently extracting building footprints from a wide range of remote sensed imagery remains a challenge due to their complex structure, variety of scales and diverse appearances. Existing convolutional neural network (CNN)-based building extraction methods are complained that they cannot detect the tiny buildings because the spatial information of CNN feature maps are lost during repeated pooling operations of the CNN, and the large buildings still have inaccurate segmentation edges. Moreover, features extracted by a CNN are always partial which restricted by the size of the respective field, and large-scale buildings with low texture are always discontinuous and holey when extracted. This paper proposes a novel multi attending path neural network (MAP-Net) for accurately extracting multiscale building footprints and precise boundaries. MAP-Net learns spatial localization-preserved multiscale features through a multi-parallel path in which each stage is gradually generated to extract high-level semantic features with fixed resolution. Then, an attention module adaptively squeezes channel-wise features from each path for optimization, and a pyramid spatial pooling module captures global dependency for refining discontinuous building footprints. Experimental results show that MAP-Net outperforms state-of-the-art (SOTA) algorithms in boundary localization accuracy as well as continuity of large buildings. Specifically, our method achieved 0.68\%, 1.74\%, 1.46\% precision, and 1.50\%, 1.53\%, 0.82\% IoU score improvement without increasing computational complexity compared with the latest HRNetv2 on the Urban 3D, Deep Globe and WHU datasets, respectively. The TensorFlow implementation is available at https://github.com/lehaifeng/MAPNet.

13 pages, 10 figures

References in corpus (8)

• Deep learning in remote sensing: a review
• DeepGlobe 2018: A Challenge to Parse the Earth through Satellite Images
• High-Resolution Representations for Labeling Pixels and Regions
• Algorithms for Semantic Segmentation of Multispectral Remote Sensing Imagery using Deep Learning
• OCNet: Object Context Network for Scene Parsing
• GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond
• Automatic Pixelwise Object Labeling for Aerial Imagery Using Stacked U-Nets
• Extracting man-made objects from remote sensing images via fast level set evolutions

Cited by in corpus (12)

• Building extraction with vision transformer
• Global and Local Contrastive Self-Supervised Learning for Semantic Segmentation of HR Remote Sensing Images
• Adversarial Shape Learning for Building Extraction in VHR Remote Sensing Images
• Self-Supervised Learning for Invariant Representations from Multi-Spectral and SAR Images
• Semi-Supervised Building Footprint Generation with Feature and Output Consistency Training
• MultiScale Probability Map guided Index Pooling with Attention-based learning for Road and Building Segmentation
• GraSS: Contrastive Learning with Gradient Guided Sampling Strategy for Remote Sensing Image Semantic Segmentation
• Building-road Collaborative Extraction from Remotely Sensed Images via Cross-Interaction
• A Novel Adaptive Deep Network for Building Footprint Segmentation
• Prompt-Driven Building Footprint Extraction in Aerial Images with Offset-Building Model
• Transport-Related Surface Detection with Machine Learning: Analyzing Temporal Trends in Madrid and Vienna
• Points2Polygons: Context-Based Segmentation from Weak Labels Using Adversarial Networks