Abstract:Deep learning methods, especially deep convolutional neural networks (DCNNs), have been widely used in high-resolution remote sensing image (HRSI) semantic segmentation. In literature, most successful DCNNs are artificially designed through a large number of experiments, which often consume lots of time and depend on rich domain knowledge. Recently, neural architecture search (NAS), as a direction for automatically designing network architectures, has achieved great success in different kinds of computer vision tasks. For HRSI semantic segmentation, NAS faces two major challenges: (1) The task's high complexity degree, which is caused by the pixel-by-pixel prediction demand in semantic segmentation, leads to a rapid expansion of the search space; (2) HRSI semantic segmentation often needs to exploit long-range dependency (i.e., a large spatial context), which means the NAS technique requires a lot of display memory in the optimization process and can be tough to converge. With the aforementioned considerations in mind, we propose a new decoupling NAS (DNAS) framework to automatically design the network architecture for HRSI semantic segmentation. In DNAS, a hierarchical search space with three levels is recommended: path-level, connection-level, and cell-level. To adapt to this hierarchical search space, we devised a new decoupling search optimization strategy to decrease the memory occupation. More specifically, the search optimization strategy consists of three stages: (1) a light super-net (i.e., the specific search space) in the path-level space is trained to get the optimal path coding; (2) we endowed the optimal path with various cross-layer connections and it is trained to obtain the connection coding; (3) the super-net, which is initialized by path coding and connection coding, is populated with kinds of concrete cell operators and the optimal cell operators are finally determined. It is worth noting that the well-designed search space can cover various network candidates and the optimization process can be done efficiently. Extensive experiments on the publicly open GID and FU datasets showed that our DNAS outperformed the state-of-the-art methods, including artificial networks and NAS methods.

Automatic Network Architecture Search for RGB-D Semantic Segmentation

ACNET: Attention Based Network to Exploit Complementary Features for RGBD Semantic Segmentation.

Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation

Hierarchical Shared Architecture Search for Real-Time Semantic Segmentation of Remote Sensing Images

DCNAS: Densely Connected Neural Architecture Search for Semantic Image Segmentation

DNAS: Decoupling Neural Architecture Search for High-Resolution Remote Sensing Image Semantic Segmentation

Nas-Hris: Automatic Design And Architecture Search Of Neural Network For Semantic Segmentation In Remote Sensing Images

Real-Time Semantic Segmentation Via Auto Depth, Downsampling Joint Decision and Feature Aggregation

Surrogate-assisted Multi-objective Neural Architecture Search for Real-time Semantic Segmentation

Spatial-information Guided Adaptive Context-aware Network for Efficient RGB-D Semantic Segmentation

M-FasterSeg: An efficient semantic segmentation network based on neural architecture search

NAS-Unet: Neural Architecture Search for Medical Image Segmentation

A Scalable Real-time Semantic Segmentation Network for Autonomous Driving

BSSNet:A Real-Time Semantic Segmentation Network for Road Scenes Inspired from AutoEncoder

SSS3D: Fast Neural Architecture Search For Efficient Three-Dimensional Semantic Segmentation

Template-Based Automatic Search of Compact Semantic Segmentation Architectures

NAS-based Recursive Stage Partial Network (RSPNet) for Light-Weight Semantic Segmentation

Graph-Guided Architecture Search for Real-Time Semantic Segmentation

Sed: Searching Enhanced Decoder with Switchable Skip Connection for Semantic Segmentation

Neural Architecture Search for Microscopy Cell Segmentation