DGCFNet: Dual Global Context Fusion Network for remote sensing image semantic segmentation

PeerJ Computer Science, Год журнала: 2025, Номер 11, С. e2786 - e2786

Опубликована: Март 27, 2025

The semantic segmentation task of remote sensing images often faces various challenges such as complex backgrounds, high inter-class similarity, and significant differences in intra-class visual attributes. Therefore, models need to capture both rich local information long-distance contextual overcome these challenges. Although convolutional neural networks (CNNs) have strong capabilities extracting information, they are limited establishing long-range dependencies due the inherent limitations convolution. While Transformer can extract through multi-head self attention mechanism, which has advantages capturing global feature dependencies. To achieve high-precision images, this article proposes a novel image network, named Dual Global Context Fusion Network (DGCFNet), is based on an encoder-decoder structure integrates CNN information. Specifically, further enhance ability modeling context, dual-branch extraction module proposed, compensation branch not only supplement but also preserve In addition, increase salient regions, cross-level interaction adopted correlation between features at different levels. Finally, optimize continuity consistency results, guided used adaptively fuse from intra layer inter layer. Extensive experiments Vaihingen, Potsdam, BLU datasets shown that proposed DGCFNet method better performance, with mIoU reaching 82.20%, 83.84% 68.87%, respectively.

Язык: Английский

A tea packaging design method based on artificial intelligence generated content

Journal of Engineering Design, Год журнала: 2025, Номер unknown, С. 1 - 33

Опубликована: Апрель 8, 2025

Язык: Английский