Redirigiendo al acceso original de articulo en 17 segundos...
Inicio  /  Applied Sciences  /  Vol: 12 Par: 7 (2022)  /  Artículo
ARTÍCULO
TITULO

Texture and Materials Image Classification Based on Wavelet Pooling Layer in CNN

Juan Manuel Fortuna-Cervantes    
Marco Tulio Ramírez-Torres    
Marcela Mejía-Carlos    
José Salomé Murguía    
José Martinez-Carranza    
Carlos Soubervielle-Montalvo and César Arturo Guerra-García    

Resumen

Convolutional Neural Networks (CNNs) have recently been proposed as a solution in texture and material classification in computer vision. However, inside CNNs, the internal layers of pooling often cause a loss of information and, therefore, is detrimental to learning the architecture. Moreover, when considering images with repetitive and essential patterns, the loss of this information affects the performance of subsequent stages, such as feature extraction and analysis. In this paper, to solve this problem, we propose a classification system with a new pooling method called Discrete Wavelet Transform Pooling (DWTP). This method is based on the image decomposition into sub-bands, in which the first level sub-band is considered as its output. The objective is to obtain approximation and detail information. As a result, this information can be concatenated in different combinations. In addition, wavelet pooling uses wavelets to reduce the size of the feature map. Combining these methods provides acceptable classification performance for three databases (CIFAR-10, DTD, and FMD). We argue that this helps eliminate overfitting and that the learning graphs reflect that the datasets show learning generalization. Therefore, our results indicate that our method based on wavelet analysis is feasible for texture and material classification. Moreover, in some cases, it outperforms traditional methods.

 Artículos similares

       
 
Spiros Paramithiotis, Maria K. Syrokou, Anastasia Papadia-Nikolaidou, Georgios Papoutsis and Eleftherios H. Drosinos    
The aim of the present study was to attempt the recreation of a highly appreciated food commodity of antiquity, called avyrtake, using information derived from ancient texts. The available information included the raw materials, the texture and the taste... ver más
Revista: Applied Sciences

 
Evgeniy V. Orekhov, Andrey Yu. Arbenin, Elena G. Zemtsova, Darya N. Sokolova, Alexandra N. Ponomareva, Maxim A. Shevtsov, Natalia M. Yudintceva and Vladimir M. Smirnov    
Modern materials science, both in terms of functional and structural materials, is actively developing towards the creation of structures with a given ordering. A wide range of methods involves ordering the structure according to a template shape. Templa... ver más
Revista: Coatings

 
Ange Nsilani Kouediatouka, Qiang Ma, Qi Liu, Fagla Jules Mawignon, Faisal Rafique and Guangneng Dong    
Surface texture is regarded as a promising solution for enhancing the tribological features of industrial materials due to its outstanding benefits, such as minimization of the contact area, enhancement of the load bearing capacity, storage of the lubric... ver más
Revista: Coatings

 
Petros Petrounias, Aikaterini Rogkala, Panagiota P. Giannakopoulou, Angeliki Christogerou, Paraskevi Lampropoulou, Spyridon Liogris, Petros Koutsovitis and Nikolaos Koukouzas    
The scope of this study focuses on the use of two different types of industrial byproducts such as slags (FeNi and Electric Arc Furnace slag) combined with natural sand as concrete aggregates as well as the evaluation of their suitability on the final ph... ver más
Revista: Applied Sciences

 
Florian Frank, Michael Tkadletz, Christoph Czettl and Nina Schalk    
As the demands for wear-resistant coatings in the cutting industry are constantly rising, new materials that have the potential to exhibit enhanced coating properties are continuously explored. Chemical vapor deposited (CVD) Zr(N,C) is a promising altern... ver más
Revista: Coatings