Redirigiendo al acceso original de articulo en 16 segundos...
Inicio  /  Algorithms  /  Vol: 15 Par: 5 (2022)  /  Artículo
ARTÍCULO
TITULO

Smart(Sampling)Augment: Optimal and Efficient Data Augmentation for Semantic Segmentation

Misgana Negassi    
Diane Wagner and Alexander Reiterer    

Resumen

Data augmentation methods enrich datasets with augmented data to improve the performance of neural networks. Recently, automated data augmentation methods have emerged, which automatically design augmentation strategies. The existing work focuses on image classification and object detection, whereas we provide the first study on semantic image segmentation and introduce two new approaches: SmartAugment and SmartSamplingAugment. SmartAugment uses Bayesian Optimization to search a rich space of augmentation strategies and achieves new state-of-the-art performance in all semantic segmentation tasks we consider. SmartSamplingAugment, a simple parameter-free approach with a fixed augmentation strategy, competes in performance with the existing resource-intensive approaches and outperforms cheap state-of-the-art data augmentation methods. Furthermore, we analyze the impact, interaction, and importance of data augmentation hyperparameters and perform ablation studies, which confirm our design choices behind SmartAugment and SmartSamplingAugment. Lastly, we will provide our source code for reproducibility and to facilitate further research.

 Artículos similares

       
 
Ting Shi and Xueyuan Peng    
Air compressors in hydrogen fuel cell vehicles play a crucial role in ensuring the stability of the cathode air system. However, they currently face challenges related to low efficiency and poor stability. To address these issues, the experimental setup ... ver más
Revista: Applied Sciences

 
Yi?an Wang, Zhe Wu and Dong Ni    
Optimizing the heliostat field aiming strategy is crucial for maximizing thermal power production in solar power tower (SPT) plants while adhering to operational constraints. Although existing approaches can yield highly optimal solutions, their consider... ver más
Revista: Applied Sciences

 
Ugur Akis and Serkan Dislitas    
In applications reliant on image processing, the management of lighting holds significance for both precise object detection and efficient energy utilization. Conventionally, lighting control involves manual switching, timed activation or automated adjus... ver más
Revista: Applied Sciences

 
Wen Gao, Yanqiang Bi, Xiyuan Li, Apeng Dong, Jing Wang and Xiaoning Yang    
Hybrid airships, combining aerodynamic lift and buoyant lift, are efficient near-space aircraft for scientific exploration, observation, and surveillance. Compared to conventional airplanes and airships, hybrid airships offer unique advantages, including... ver más
Revista: Aerospace

 
Juan Francisco Mora-Sánchez, Josep Ribes, Josué González-Camejo, Aurora Seco and María Victoria Ruano    
This research lays a foundation for optimised membrane photobioreactor performance and introduces novel control parameters crucial for advancing microalgae cultivation techniques and promoting environmental sustainability. Particularly, this study presen... ver más
Revista: Water