Unknow Classs Classification using Deep Learning Technique

Traditional Fruits feature extraction methods focus on spectral features and neglect spatial features, its extraction method is set in advance and is not suitable for all Fruits images. Faced with these problems, we propose a three-dimensional convolutional network for Fruits classification, which consists of a convolutional layer,2 down sampling layers, 2 identification layers, a flatten layer, and 4 fully connected layers. The proposed network employs three-dimensional convolution operation to extract spectral-spatial features from Fruits images, there are two reasons for this, the first reason is three-dimensional convolution can automatically learn a large number of mappings between input and output. The second reason is three-dimensional convolution can effectively extract spectral-spatial features and improve network classification performance. In order to extract high-level features and prevent network performance degradation, the proposed network adopts residual connections. More importantly, the OpenMax algorithm is employed to detect Fruits unknown targets. In addition to the probability that the output belongs to a known class, the OpenmMax adds the probability that the predicted input belongs to unknown classes, as a result, the deep convolutional network can respond to inputs of unknown classes. Experiments based on typical Fruits data show that the proposed network perform accurately in the known classes’ classification and the openmax algorithm is suitable for unknown targets detection of Fruits images.

Known Classes Classification And Unkown Classes Targets Detection Of

FruitsImages Based On Convolutional Neural Networks

Lixiong Zhang,Yuanxi Peng,Tian Jiang ,Yu Liu,Kecheng Gong,Lu Liu,Liyuan Zhao,Longlong Zhang

(National University of Defense and Technology,College of Computer,CHANG’sha 410073,China)

Abstract

Traditional Fruitsfeature extraction methods focus on spectral features and neglect spatial features,its extraction method

is set in advance and is not suitable for all Fruitsimages. Faced with these problems, we propose a three-dimensional

convolutional network for Fruitsclassification, which consists of a convolutional layer,2 downsampling layers, 2

identification layers, a flatten layer, and 4 fully connected layers. The proposed network employs three-dimensional convolution

operation to extract spectral-spatial features from Fruitsimages,there are two reasons for this, the first reason is

three-dimensional convolution can automatically learn a large number of mappings between input and output.The second reason is

three-dimensional convolution can effectively extract spectral-spatial features and improve network classification performance. In

order to extract high-level features and prevent network performance degradation, the proposed network adopts residual

connections.More importantly, the OpenMax algorithm is employed to detect Fruitsunknown targets. In addition to the

probability that the output belongs to a known class, the OpenmMx adds the probability that the predicted input belongs to unknown

classes, as a result,the deep convolutional network can respond to inputs of unknown classes.experiments based on typical

Fruitsdata show that the proposed network perform accurately in the known classes classification and the openmax algorithm

is suitable for unknown targets detection of Fruitsimages.

Keywords:FruitsImages ,Three-dimensional convolution, Spectral-spatial features, OpenMax,Known class

classification,Unknown classes detection,Residual connections, Extreme value theory

Known Classes Classification And Unkown Classes Targets Detection Of

FruitsImages Based On Convolutional Neural Networks

Lixiong Zhang,Yuanxi Peng,Tian Jiang ,Yu Liu,Kecheng Gong,Lu Liu,Liyuan Zhao,Longlong Zhang

(National University of Defense and Technology,College of Computer,CHANG’sha 410073,China)

Abstract

Traditional Fruitsfeature extraction methods focus on spectral features and neglect spatial features,its extraction method

is set in advance and is not suitable for all Fruitsimages. Faced with these problems, we propose a three-dimensional

convolutional network for Fruitsclassification, which consists of a convolutional layer,2 downsampling layers, 2

identification layers, a flatten layer, and 4 fully connected layers. The proposed network employs three-dimensional convolution

operation to extract spectral-spatial features from Fruitsimages,there are two reasons for this, the first reason is

three-dimensional convolution can automatically learn a large number of mappings between input and output.The second reason is

three-dimensional convolution can effectively extract spectral-spatial features and improve network classification performance. In

order to extract high-level features and prevent network performance degradation, the proposed network adopts residual

connections.More importantly, the OpenMax algorithm is employed to detect Fruitsunknown targets. In addition to the

probability that the output belongs to a known class, the OpenmMx adds the probability that the predicted input belongs to unknown

classes, as a result,the deep convolutional network can respond to inputs of unknown classes.experiments based on typical

Fruitsdata show that the proposed network perform accurately in the known classes classification and the openmax algorithm

is suitable for unknown targets detection of Fruitsimages.

Keywords:FruitsImages ,Three-dimensional convolution, Spectral-spatial features, OpenMax,Known class

classification,Unknown classes detection,Residual connections, Extreme value theory

Keywords:FruitsImages ,Three-dimensional convolution, Spectral-spatial features, OpenMax,Known class

classification,Unknown classes detection,Residual connections, Extreme value theory