The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Visando as diferentes habilidades dos algoritmos de desembaçamento em diferentes concentrações de neblina, este artigo propõe um algoritmo de classificação de imagens de neblina para um conjunto de dados de amostra pequeno e desequilibrado baseado em uma rede neural de convolução, que pode classificar as imagens de neblina antecipadamente, de modo a melhorar o efeito e capacidade adaptativa do algoritmo de desembaçamento de imagem em condições de neblina e neblina. A fim de resolver os problemas de interferência ambiental, interferência de profundidade de campo da câmera e distribuição desigual de recursos em imagens de neblina, o método de aumento de dados CutBlur-Gauss e estratégias de perda focal e suavização de rótulo são usados para melhorar a precisão da classificação. Ele é comparado com o algoritmo de aprendizado de máquina SVM e os algoritmos clássicos de classificação de redes neurais de convolução alexnet, resnet34, resnet50 e resnet101. Este algoritmo atinge 94.5% de precisão de classificação no conjunto de dados deste artigo, o que excede outros excelentes algoritmos de comparação atualmente e atinge a melhor precisão. Está provado que o algoritmo aprimorado possui melhor precisão de classificação.
Fuxiang LIU
Beijing Institute of Technology
Chen ZANG
Beijing Institute of Technology
Lei LI
Science and Technology on Avionics Integration Laboratory
Chunfeng XU
Beijing Institute of Technology
Jingmin LUO
CHINAROCKERT CO., LTD.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Fuxiang LIU, Chen ZANG, Lei LI, Chunfeng XU, Jingmin LUO, "ConvNeXt-Haze: A Fog Image Classification Algorithm for Small and Imbalanced Sample Dataset Based on Convolutional Neural Network" in IEICE TRANSACTIONS on Information,
vol. E106-D, no. 4, pp. 488-494, April 2023, doi: 10.1587/transinf.2022IIP0009.
Abstract: Aiming at the different abilities of the defogging algorithms in different fog concentrations, this paper proposes a fog image classification algorithm for a small and imbalanced sample dataset based on a convolution neural network, which can classify the fog images in advance, so as to improve the effect and adaptive ability of image defogging algorithm in fog and haze weather. In order to solve the problems of environmental interference, camera depth of field interference and uneven feature distribution in fog images, the CutBlur-Gauss data augmentation method and focal loss and label smoothing strategies are used to improve the accuracy of classification. It is compared with the machine learning algorithm SVM and classical convolution neural network classification algorithms alexnet, resnet34, resnet50 and resnet101. This algorithm achieves 94.5% classification accuracy on the dataset in this paper, which exceeds other excellent comparison algorithms at present, and achieves the best accuracy. It is proved that the improved algorithm has better classification accuracy.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2022IIP0009/_p
Copiar
@ARTICLE{e106-d_4_488,
author={Fuxiang LIU, Chen ZANG, Lei LI, Chunfeng XU, Jingmin LUO, },
journal={IEICE TRANSACTIONS on Information},
title={ConvNeXt-Haze: A Fog Image Classification Algorithm for Small and Imbalanced Sample Dataset Based on Convolutional Neural Network},
year={2023},
volume={E106-D},
number={4},
pages={488-494},
abstract={Aiming at the different abilities of the defogging algorithms in different fog concentrations, this paper proposes a fog image classification algorithm for a small and imbalanced sample dataset based on a convolution neural network, which can classify the fog images in advance, so as to improve the effect and adaptive ability of image defogging algorithm in fog and haze weather. In order to solve the problems of environmental interference, camera depth of field interference and uneven feature distribution in fog images, the CutBlur-Gauss data augmentation method and focal loss and label smoothing strategies are used to improve the accuracy of classification. It is compared with the machine learning algorithm SVM and classical convolution neural network classification algorithms alexnet, resnet34, resnet50 and resnet101. This algorithm achieves 94.5% classification accuracy on the dataset in this paper, which exceeds other excellent comparison algorithms at present, and achieves the best accuracy. It is proved that the improved algorithm has better classification accuracy.},
keywords={},
doi={10.1587/transinf.2022IIP0009},
ISSN={1745-1361},
month={April},}
Copiar
TY - JOUR
TI - ConvNeXt-Haze: A Fog Image Classification Algorithm for Small and Imbalanced Sample Dataset Based on Convolutional Neural Network
T2 - IEICE TRANSACTIONS on Information
SP - 488
EP - 494
AU - Fuxiang LIU
AU - Chen ZANG
AU - Lei LI
AU - Chunfeng XU
AU - Jingmin LUO
PY - 2023
DO - 10.1587/transinf.2022IIP0009
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E106-D
IS - 4
JA - IEICE TRANSACTIONS on Information
Y1 - April 2023
AB - Aiming at the different abilities of the defogging algorithms in different fog concentrations, this paper proposes a fog image classification algorithm for a small and imbalanced sample dataset based on a convolution neural network, which can classify the fog images in advance, so as to improve the effect and adaptive ability of image defogging algorithm in fog and haze weather. In order to solve the problems of environmental interference, camera depth of field interference and uneven feature distribution in fog images, the CutBlur-Gauss data augmentation method and focal loss and label smoothing strategies are used to improve the accuracy of classification. It is compared with the machine learning algorithm SVM and classical convolution neural network classification algorithms alexnet, resnet34, resnet50 and resnet101. This algorithm achieves 94.5% classification accuracy on the dataset in this paper, which exceeds other excellent comparison algorithms at present, and achieves the best accuracy. It is proved that the improved algorithm has better classification accuracy.
ER -