scholarly journals Exploring Misclassification Information for Fine-Grained Image Classification

Sensors ◽  
2021 ◽  
Vol 21 (12) ◽  
pp. 4176
Author(s):  
Da-Han Wang ◽  
Wei Zhou ◽  
Jianmin Li ◽  
Yun Wu ◽  
Shunzhi Zhu
Keyword(s):  
Image Classification ◽  
Classification Accuracy ◽  
Experimental Results ◽  
Classification Method ◽  
Public Image ◽  
Classification Models ◽  
Classification Methods ◽  
Fine Grained ◽  
Number Of Classes ◽  
Image Datasets

Fine-grained image classification is a hot topic that has been widely studied recently. Many fine-grained image classification methods ignore misclassification information, which is important to improve classification accuracy. To make use of misclassification information, in this paper, we propose a novel fine-grained image classification method by exploring the misclassification information (FGMI) of prelearned models. For each class, we harvest the confusion information from several prelearned fine-grained image classification models. For one particular class, we select a number of classes which are likely to be misclassified with this class. The images of selected classes are then used to train classifiers. In this way, we can reduce the influence of irrelevant images to some extent. We use the misclassification information for all the classes by training a number of confusion classifiers. The outputs of these trained classifiers are combined to represent images and produce classifications. To evaluate the effectiveness of the proposed FGMI method, we conduct fine-grained classification experiments on several public image datasets. Experimental results prove the usefulness of the proposed method.

scholarly journals Fine-Grained Image Classification for Crop Disease Based on Attention Mechanism

2020 ◽  
Vol 11 ◽  
Author(s):  
Guofeng Yang ◽  
Yong He ◽  
Yong Yang ◽  
Beibei Xu
Keyword(s):  
Image Classification ◽  
Classification Accuracy ◽  
Attention Mechanism ◽  
Classification Model ◽  
Classification Models ◽  
Fine Grained ◽  
Complex Scenes ◽  
Crop Disease ◽  
Visual Disturbances

Fine-grained image classification is a challenging task because of the difficulty in identifying discriminant features, it is not easy to find the subtle features that fully represent the object. In the fine-grained classification of crop disease, visual disturbances such as light, fog, overlap, and jitter are frequently encountered. To explore the influence of the features of crop leaf images on the classification results, a classification model should focus on the more discriminative regions of the image while improving the classification accuracy of the model in complex scenes. This paper proposes a novel attention mechanism that effectively utilizes the informative regions of an image, and describes the use of transfer learning to quickly construct several fine-grained image classification models of crop disease based on this attention mechanism. This study uses 58,200 crop leaf images as a dataset, including 14 different crops and 37 different categories of healthy/diseased crops. Among them, different diseases of the same crop have strong similarities. The NASNetLarge fine-grained classification model based on the proposed attention mechanism achieves the best classification effect, with an F1 score of up to 93.05%. The results show that the proposed attention mechanism effectively improves the fine-grained classification of crop disease images.

scholarly journals Generating Natural Counterfactual Visual Explanations

2020 ◽  
Author(s):  
Wenqi Zhao ◽  
Satoshi Oyama ◽  
Masahito Kurihara
Keyword(s):  
Image Classification ◽  
Classification Accuracy ◽  
Classification Task ◽  
Learning Models ◽  
Generative Adversarial Network ◽  
Fine Grained ◽  
Class A ◽  
Adversarial Network ◽  
Class B ◽  
Machine Learning Models

Counterfactual explanations help users to understand the behaviors of machine learning models by changing the inputs for the existing outputs. For an image classification task, an example counterfactual visual explanation explains: "for an example that belongs to class A, what changes do we need to make to the input so that the output is more inclined to class B." Our research considers changing the attribute description text of class A on the basis of the attributes of class B and generating counterfactual images on the basis of the modified text. We can use the prediction results of the model on counterfactual images to find the attributes that have the greatest effect when the model is predicting classes A and B. We applied our method to a fine-grained image classification dataset and used the generative adversarial network to generate natural counterfactual visual explanations. To evaluate these explanations, we used them to assist crowdsourcing workers in an image classification task. We found that, within a specific range, they improved classification accuracy.

A New Network Traffic Classification Method Based on Optimized Hadamard Matrix and ECOC-SVM

2014 ◽  
Vol 989-994 ◽  
pp. 1895-1900
Author(s):  
Hong Zhi Wang ◽  
Li Hui Yan
Keyword(s):  
Network Traffic ◽  
Classification Accuracy ◽  
Hadamard Matrix ◽  
Classification Method ◽  
Traffic Classification ◽  
Classification Methods ◽  
Computational Burden ◽  
Network Traffic Classification ◽  
Encrypted Traffic

The traditional network traffic classification methods have many shortcomings, the classification accuracy is not high, the encrypted traffic cannot be analyzed, and the computational burden is usually large. To overcome above problems, this paper presents a new network traffic classification method based on optimized Hadamard matrix and ECOC. Through restructuring the Hadamard matrix and erasing the interference rows and columns, the ECOC table is optimized while eliminating SVM sample imbalance, and the error correcting ability for classification is reserved. The experiments results show that the proposed method outperform in network traffic classification and improve the classification accuracy.

scholarly journals PERBAIKAN METODE IDENTIFIKASI POTENSI PENGEMBANGAN LAHAN UNTUK TAMBAK AIR PAYAU SISTEM EKSTENSIF LEWAT INTEGRASI LOGIKA SAMAR DAN PENGINDERAAN JAUH

2016 ◽  
Vol 5 (2) ◽  
pp. 317
Author(s):  
Tarunamulia Tarunamulia ◽  
Jesmond Sammut ◽  
Akhmad Mustafa
Keyword(s):  
Fuzzy Logic ◽  
Maximum Likelihood ◽  
Image Classification ◽  
Supervised Classification ◽  
Training Data ◽  
Classification Method ◽  
Classification Methods ◽  
Elevation Data ◽  
Landsat 7 ◽  
Brackishwater Pond

Tersedianya data potensi lahan tambak yang cepat, akurat dan lengkap untuk kebutuhan pengelolaan kawasan pengembangan perikanan budidaya air payau harus didukung oleh metode identifikasi yang efektif dan efisien. Penelitian ini bertujuan untuk mengupayakan peningkatan kualitas metode klasifikasi multispektral dalam penginderaan jauh dalam mengidentifikasi potensi lahan tambak ekstensif dengan mengintegrasikan logika samar dalam proses klasifikasi citra. Citra landsat-7 ETM+ (30 m), data elevasi digital dan data pengecekan lapang untuk wilayah pantai (kawasan tambak ekstensif/tradisional) Kecamatan Kembang Tanjung, Pidie, Nangroe Aceh Darussalam (NAD) digunakan sebagai bahan utama dalam penelitian ini. Klasifikasi multispektral standar secara terbimbing diperbaiki melalui pengambilan data training secara cermat, yang diikuti dengan uji keterpisahan objek, pemrosesan pasca-klasifikasi dan analisis tingkat ketelitian. Hasil klasifikasi dengan tingkat ketelitian terbaik dari berbagai algoritma yang diujikan untuk tiga saluran selanjutnya dibandingkan dengan hasil klasifikasi dengan menggunakan logika samar. Dari hasil penelitian diketahui bahwa klasifikasi multispektral standar dengan algoritma Maximum Likelihood mampu menghasilkan informasi penutup lahan yang cukup lengkap dan rinci pada wilayah pertambakan dengan ketelitian yang cukup baik (>86%). Tingkat ketelitian yang sama juga masih dijumpai walaupun hanya melibatkan kombinasi 3 saluran terbaik (5,4, dan 3) yang dipilih berdasarkan analisis statistik nilai kecerahan piksel. Dengan membandingkan hasil terbaik dari metode klasifikasi standar yang berbasis logika biner (boolean) dengan hasil klasifikasi citra dengan logika samar dalam pengklasifikasian wilayah tambak, diketahui bahwa klasifikasi citra dengan logika samar mampu memperlihatkan hasil klasifikasi yang sangat baik untuk menentukan batas wilayah tambak yang tidak bisa dilakukan secara langsung bahkan oleh metode standar dengan algoritma terbaik. Dan dengan penambahan satu variabel kunci untuk tambak ekstensif seperti elevasi dalam klasifikasi, klasifikasi dengan logika samar dapat digunakan untuk memprediksi potensi pengembangan lahan budidaya tambak ekstensif dan kemungkinan tumpang tindih dengan penggunaan lahan lainnya.The availability of immediate, accurate and complete data on potential pond area as a baseline data for land management of brackishwater aquaculture must be supported by effective and efficient identification methods. The objective of this study was to explore the possibility of improving the quality of multispectral image classification methods in identifying potential areas for extensive brackishwater aquaculture through the integration of fuzzy logic and classification of remotely sensed data. 2002 Landsat-7 Enhanced Thematic Mapper Data (30-m pixels), digital elevation data, and groundtruthing of training data (region of interest/ROI) of Kembang Tanjung coastal areas (Pidie, NAD) were used as the primary data in this study. Standard supervised multispectral classification methods were enhanced by collecting appropriate and unbiased training data, applying separability measures of ROI pairs, employing post-classification analysis, and assessing the accuracy of classification results. Different types of standard supervised classification algorithms were evaluated and a classification output with the highest accuracy was selected to be compared with the result from fuzzy logic classification. The study showed that a supervised classification method based on maximum likelihood analysis produced the best classification output of land use-cover over the coastal region (overall accuracy > 86%). The accuracy remained at the same level although it involved only the best composite of 3 bands (5,4, and 3) determined by a rigorous statistical analysis of brightness values of pixels. It was clear that the fuzzy-based classification method was more effective in identifying potential extensive brackishwater pond areas compared to the best standard image classification based on binary logic (maximum likelihood). Also, by integrating elevation data as another key variable to determine the suitability of land for extensive brackishwater aquaculture, the fuzzy classification can be used to more accurately predict potential area suited for brackishwater aquaculture ponds and any possible overlapping activity with other land uses.

scholarly journals EFFECT OF PANSHARPENED IMAGE ON SOME OF PIXEL BASED AND OBJECT BASED CLASSIFICATION ACCURACY

2016 ◽  
Vol XLI-B7 ◽  
pp. 235-239 ◽  
Author(s):  
P. Karakus ◽  
H. Karabork
Keyword(s):  
Spatial Resolution ◽  
Spectral Resolution ◽  
Classification Accuracy ◽  
Multispectral Image ◽  
Classification Method ◽  
Support Vector ◽  
Classification Methods ◽  
Panchromatic Image ◽  
Object Based ◽  
Spot 5

Classification is the most important method to determine type of crop contained in a region for agricultural planning. There are two types of the classification. First is pixel based and the other is object based classification method. While pixel based classification methods are based on the information in each pixel, object based classification method is based on objects or image objects that formed by the combination of information from a set of similar pixels. Multispectral image contains a higher degree of spectral resolution than a panchromatic image. Panchromatic image have a higher spatial resolution than a multispectral image. Pan sharpening is a process of merging high spatial resolution panchromatic and high spectral resolution multispectral imagery to create a single high resolution color image. The aim of the study was to compare the potential classification accuracy provided by pan sharpened image. In this study, SPOT 5 image was used dated April 2013. 5m panchromatic image and 10m multispectral image are pan sharpened. Four different classification methods were investigated: maximum likelihood, decision tree, support vector machine at the pixel level and object based classification methods. SPOT 5 pan sharpened image was used to classification sun flowers and corn in a study site located at Kadirli region on Osmaniye in Turkey. The effects of pan sharpened image on classification results were also examined. Accuracy assessment showed that the object based classification resulted in the better overall accuracy values than the others. The results that indicate that these classification methods can be used for identifying sun flower and corn and estimating crop areas.

Benchmarking Classification Models for Cell Viability on Novel Cancer Image Datasets

2019 ◽  
Vol 14 (2) ◽  
pp. 108-114 ◽  
Author(s):  
Akın Özkan ◽  
Sultan Belgin İşgör ◽  
Gökhan Şengül ◽  
Yasemin Gülgün İşgör
Keyword(s):  
Decision Making ◽  
Cell Culture ◽  
Cell Viability ◽  
Cell Biology ◽  
Experimental Results ◽  
Myelogenous Leukemia ◽  
Classification Models ◽  
K Nearest Neighbor ◽  
Viability Analysis ◽  
Image Datasets

Background: Dye-exclusion based cell viability analysis has been broadly used in cell biology including anticancer drug discovery studies. Viability analysis refers to the whole decision making process for the distinction of dead cells from live ones. Basically, cell culture samples are dyed with a special stain called trypan blue, so that the dead cells are selectively colored to darkish. This distinction provides critical information that may be used to expose influences of the studied drug on considering cell culture including cancer. Examiner’s experience and tiredness substantially affect the consistency throughout the manual observation of cell viability. The unsteady results of cell viability may end up with biased experimental results accordingly. Therefore, a machine learning based automated decision-making procedure is inevitably needed to improve consistency of the cell viability analysis. Objective: In this study, we investigate various combinations of classifiers and feature extractors (i.e. classification models) to maximize the performance of computer vision-based viability analysis. Method: The classification models are tested on novel hemocytometer image datasets which contain two types of cancer cell images, namely, caucasian promyelocytic leukemia (HL60), and chronic myelogenous leukemia (K562). Results: From the experimental results, k-Nearest Neighbor (KNN) and Random Forest (RF) by combining Local Phase Quantization (LPQ) achieve the lowest misclassification rates that are 0.031 and 0.082, respectively. Conclusion: The experimental results show that KNN and RF with LPQ can be powerful alternatives to the conventional manual cell viability analysis. Also, the collected datasets are released from the “biochem.atilim.edu.tr/datasets/” web address publically to academic studies.

A CONVblock for Convolutional Neural Networks

2021 ◽  
pp. 100-113
Author(s):  
Hmidi Alaeddine ◽  
Malek Jihene
Keyword(s):  
Neural Networks ◽  
Image Classification ◽  
Convolutional Neural Networks ◽  
Experimental Results ◽  
Classification Models ◽  
Deep Architecture ◽  
Improved Performance ◽  
Convolution Filters ◽  
Classification Database

The reduction in the size of convolution filters has been shown to be effective in image classification models. They make it possible to reduce the calculation and the number of parameters used in the operations of the convolution layer while increasing the efficiency of the representation. The authors present a deep architecture for classification with improved performance. The main objective of this architecture is to improve the main performances of the network thanks to a new design based on CONVblock. The proposal is evaluated on a classification database: CIFAR-10 and MNIST. The experimental results demonstrate the effectiveness of the proposed method. This architecture offers an error of 1.4% on CIFAR-10 and 0.055% on MNIST.

scholarly journals Research of the efficiency of application of technologies of wavelet analysis in the tasks of recognition of images

2021 ◽  
Vol 1 (1) ◽  
pp. 23-28
Author(s):  
V. S. Maraev
Keyword(s):  
Wavelet Analysis ◽  
Classification Accuracy ◽  
Experimental Comparison ◽  
Support Vector ◽  
Classification Models ◽  
Classification Methods ◽  
Test Results ◽  
Bayes Classifier ◽  
Fractal Method ◽  
Vector Machines

The work is devoted to the experimental comparison of the accuracy of classification methods on the problem of pattern recognition in images using wavelet analysis technologies and without. In particular, the interaction of the ring-projection wavelet-fractal method for identifying features with classical classification methods such as "Naive Bayes classifier" and "Support vector machines" is investigated. The experimental test results are given in the form of a table. As a result, it is established that the introduction of wavelet analysis into the construction of image classification models is justified, and leads to a relatively small but significant increase in the classification accuracy.

Image Classification Based on Sparse Representation and ROI

2014 ◽  
Vol 556-562 ◽  
pp. 4906-4910
Author(s):  
Hui Hui Zhao ◽  
Jun Ding Sun
Keyword(s):  
Sparse Representation ◽  
Image Classification ◽  
Extraction Methods ◽  
Saliency Map ◽  
Regions Of Interest ◽  
Experimental Results ◽  
Classification Method ◽  
Saliency Maps ◽  
Roi Extraction

A new image classification method based on regions of interest (ROI) and sparse representation is introduced in the paper. Firstly, the saliency map of each image is extracted by different methods. Then, we choose sparse representation to represent and classify the saliency maps. Four different ROI extraction methods are chosen as examples to evaluate the performance of the proposed method. Experimental results show that it is more effective for image classification based on ROI.

scholarly journals Convolutional Attention Network with Maximizing Mutual Information for Fine-Grained Image Classification

Symmetry ◽  
2020 ◽  
Vol 12 (9) ◽  
pp. 1511
Author(s):  
Fenglei Wang ◽  
Hao Zhou ◽  
Shuohao Li ◽  
Jun Lei ◽  
Jun Zhang
Keyword(s):  
Mutual Information ◽  
Image Classification ◽  
Semantic Features ◽  
Classification Methods ◽  
Global Features ◽  
Fine Grained ◽  
Symmetric Structure ◽  
Learning Techniques ◽  
Local And Global Features ◽  
Image Pairs

Fine-grained image classification has seen a great improvement benefiting from the advantages of deep learning techniques. Most fine-grained image classification methods focus on extracting discriminative features and combining the global features with the local ones. However, the accuracy is limited due to the inter-class similarity and the inner-class divergence as well as the lack of enough labelled images to train a deep network which can generalize to fine-grained classes. To deal with these problems, we develop an algorithm which combines Maximizing the Mutual Information (MMI) with the Learning Attention (LA). We make use of MMI to distill knowledge from the image pairs which contain the same object. Meanwhile we take advantage of the LA mechanism to find the salient region of the image to enhance the information distillation. Our model can extract more discriminative semantic features and improve the performance on fine-grained image classification. Our model has a symmetric structure, in which the paired images are inputted into the same network to extract the local and global features for the subsequent MMI and LA modules. We train the model by maximizing the mutual information and minimizing the cross-entropy stage by stage alternatively. Experiments show that our model can improve the performance of the fine-grained image classification effectively.

Sign in / Sign up

Export Citation Format

Share Document