Redirigiendo al acceso original de articulo en 15 segundos...
Inicio  /  Applied Sciences  /  Vol: 13 Par: 5 (2023)  /  Artículo
ARTÍCULO
TITULO

GhostNeXt: Rethinking Module Configurations for Efficient Model Design

Kiseong Hong    
Gyeong-hyeon Kim and Eunwoo Kim    

Resumen

Despite the continuous development of convolutional neural networks, it remains a challenge to achieve performance improvement with fewer parameters and floating point operations (FLOPs) as a light-weight model. In particular, excessive expressive power on a module is a crucial cause of skyrocketing the computational cost of the entire network. We argue that it is necessary to optimize the entire network by optimizing single modules or blocks of the network. Therefore, we propose GhostNeXt, a promising alternative to GhostNet, by adjusting the module configuration inside the Ghost block. We introduce a controller to select channel operations of the module dynamically. It holds a plug-and-play component that is more useful than the existing approach. Experiments on several classification tasks demonstrate that the proposed method is a better alternative to convolution layers in baseline models. GhostNeXt achieves competitive recognition performance compared to GhostNet and other popular models while reducing computational costs on the benchmark datasets.

 Artículos similares

       
 
Juan Jin, Seunghan Kim and Jiwon Moon    
Korean forests are highly vulnerable to forest fires, which can severely damage property and human life. This necessitates the establishment of a rapid response system and the construction of firebreaks to prevent the spread of fires and protect key faci... ver más
Revista: Applied Sciences

 
Yuhuan Wu and Yonghong Wu    
Salient object detection (SOD) aims to identify the most visually striking objects in a scene, simulating the function of the biological visual attention system. The attention mechanism in deep learning is commonly used as an enhancement strategy which e... ver más
Revista: Algorithms

 
Xiaodong Cui, Zhuofan He, Yangtao Xue, Keke Tang, Peican Zhu and Jing Han    
Underwater Acoustic Target Recognition (UATR) plays a crucial role in underwater detection devices. However, due to the difficulty and high cost of collecting data in the underwater environment, UATR still faces the problem of small datasets. Few-shot le... ver más

 
Sergio Bonaccorsi, Marco Felice Montaruli, Pierluigi Di Lizia, Moreno Peroni, Alessandro Panico, Marco Rigamonti and Francesco Del Prete    
The increasing number of objects in Earth orbit has encouraged the development of space surveillance and tracking (SST) applications. A critical aspect of SST is the identification and characterization of close encounters between pairs of space objects. ... ver más
Revista: Aerospace

 
Xuedong Li, Yuan Xie, Yumo Tian and Fengjiang An    
Extravehicular activity (EVA) is a key point and a difficult point for manned spaceflight tasks, as well as an inevitable trend in the development of the manned spaceflight industry. Equipment maintenance, load installation, and extravehicular routing in... ver más
Revista: Aerospace