English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recently, a massive number of deep learning based approaches have beensuccessfully applied to various remote sensing image (RSI) recognition tasks.However, most existing advances of deep learning methods in the RSI fieldheavily rely on the features extracted by the manually designed backbonenetwork, which severely hinders the potential of deep learning models due thecomplexity of RSI and the limitation of prior knowledge. In this paper, weresearch a new design paradigm for the backbone architecture in RSI recognitiontasks, including scene classification, land-cover classification and objectdetection. A novel one-shot architecture search framework based onweight-sharing strategy and evolutionary algorithm is proposed, called RSBNet,which consists of three stages: Firstly, a supernet constructed in a layer-wisesearch space is pretrained on a self-assembled large-scale RSI dataset based onan ensemble single-path training strategy. Next, the pre-trained supernet isequipped with different recognition heads through the switchable recognitionmodule and respectively fine-tuned on the target dataset to obtaintask-specific supernet. Finally, we search the optimal backbone architecturefor different recognition tasks based on the evolutionary algorithm without anynetwork training. Extensive experiments have been conducted on five benchmarkdatasets for different recognition tasks, the results show the effectiveness ofthe proposed search paradigm and demonstrate that the searched backbone is ableto flexibly adapt different RSI recognition tasks and achieve impressiveperformance.

RSBNet: One-Shot Neural Architecture Search for A Backbone Network in Remote Sensing Image Recognition