Image Classification Results Research Articles

Interpretability-Guided Inductive Bias For Deep Learning Based Medical Image.

Deep learning methods provide state of the art performance for supervised learning based medical image analysis. However it is essential that trained models extract clinically relevant features for downstream tasks as, otherwise, shortcut learning and generalization issues can occur. Furthermore in the medical field, trustability and transparency of current deep learning systems is a much desired property. In this paper we propose an interpretability-guided inductive bias approach enforcing that learned features yield more distinctive and spatially consistent saliency maps for different class labels of trained models, leading to improved model performance. We achieve our objectives by incorporating a class-distinctiveness loss and a spatial-consistency regularization loss term. Experimental results for medical image classification and segmentation tasks show our proposed approach outperforms conventional methods, while yielding saliency maps in higher agreement with clinical experts. Additionally, we show how information from unlabeled images can be used to further boost performance. In summary, the proposed approach is modular, applicable to existing network architectures used for medical imaging applications, and yields improved learning rates, model robustness, and model interpretability.

RCMNet: A deep learning model assists CAR-T therapy for leukemia

Acute leukemia is a type of blood cancer with a high mortality rate. Current therapeutic methods include bone marrow transplantation, supportive therapy, and chemotherapy. Although a satisfactory remission of the disease can be achieved, the risk of recurrence is still high. Therefore, novel treatments are demanding. Chimeric antigen receptor-T (CAR-T) therapy has emerged as a promising approach to treating and curing acute leukemia. To harness the therapeutic potential of CAR-T cell therapy for blood diseases, reliable cell morphological identification is crucial. Nevertheless, the identification of CAR-T cells is a big challenge posed by their phenotypic similarity with other blood cells. To address this substantial clinical challenge, herein we first construct a CAR-T dataset with 500 original microscopy images after staining. Following that, we create a novel integrated model called RCMNet (ResNet18 with Convolutional Block Attention Module and Multi-Head Self-Attention) that combines the convolutional neural network (CNN) and Transformer. The model shows 99.63% top-1 accuracy on the public dataset. Compared with previous reports, our model obtains satisfactory results for image classification. Although testing on the CAR-T cell dataset, a decent performance is observed, which is attributed to the limited size of the dataset. Transfer learning is adapted for RCMNet and a maximum of 83.36% accuracy is achieved, which is higher than that of other state-of-the-art models. This study evaluates the effectiveness of RCMNet on a big public dataset and translates it to a clinical dataset for diagnostic applications.

Combating spatial redundancy with spectral norm attention in convolutional learners

There is an inherent and longstanding challenge for vision learners to exploit informative features from digital images with spatial redundancy. Given pre-processing image methods require task-specific customization and may rise unanticipated poor performance due to redundancy removal, we explore improving vision learners to combat spatial redundancy during vision learning, a task-agnostic and robust solution. Among popular vision learners, vision transformers with self-attention can mitigate pixel redundancy by capturing global dependencies, while convolutional learners fall into locality via a limited receptive field. To this end, based on investigating inter-pixel spatial redundancy of images, in this work, we propose spectral norm attention (SNA), a novel yet efficient attention block to help convolutional neural networks (CNNs) highlight informative features. We can seamlessly plug SNA into off-the-shelf CNNs to suppress the contributions of redundant features by globally differentiating and weighting. In particular, SNA performs singular value decomposition (SVD) on intermediate features of each image within a mini-batch to obtain its spectral norm. The features in the direction of the spectral norm are most informative, while the discriminative features in other directions leave less. Hence, we apply the rank-one approximation of the spectral norm direction as attention weights to enhance informative features. Besides, we adopt the power iteration algorithm to approximate the spectral norm to significantly reduce the matrix computation overhead during training, thus keeping inference speed on par with vanilla CNNs. We extensively evaluate our SNA on four mainstream natural datasets to demonstrate the effectiveness and favourability of our SNA against its counterparts. In addition, the experimental results of image classification and object detection show our SNA can bring more gains to medical images with heavy redundancy than other state-of-the-art attention modules.

Application of the wavelet transformation theory in the algorithm for constructing a quasigeoid model

Purpose. To investigate the interaction of geodesic and normal altitude indicators according to quasigeoid data, the joint use of space measurements and those performed on the Earths surface in the implementation of geodetic tasks. In this article, the task is to create a calculation algorithm for further research on the quasigeoid model and the application of the model in solving geodetic problems. Methodology. Reliable determination of the height anomaly requires great accuracy, therefore, the theory of wavelet-transformation was used in the model of the variant of space technologies as an alternative to the laborious leveling of the Earths surface, which characterizes the actual fluctuations from the normal of the Earths gravitational field, when calculating the mean square deviations of the plumb line is an urgent task. Findings. A block diagram of the calculation algorithm has been compiled using a software package to solve the boundary problem of physical geodesy, in which the Earths surface is subject to modern space measurements. Originality. The use of wavelet analysis for processing information from satellite data in geodesy improves the results of image classification, and the coefficients of the wavelet transformation can be used as indicators for recognizing the coordinates of points with high accuracy. Practical value. Application of the theory of wavelet transformations as a powerful mathematical tool for solving problems of geodetic information, data compression and recovery, increasing computing performance, encoding information.

Image Classification Learning Method Incorporating Zero-Sample Learning and Small-Sample Learning

At present, artificial intelligence algorithms based on deep learning have achieved good results in image classification, biometric recognition, medical diagnosis, and other fields. However, in practice, many times researchers are unable to obtain a large number of samples due to many limitations or high sampling costs. Therefore, image sorting zero-sampling order research algorithms have become the central engine of intelligent processing and a hot spot for current research. Because of the need for the development of deep learning prediction capability, coupled with the emergence of time and technical-level drawbacks, the advantages of zero-sample and small-sample are gradually emerging, so this paper chooses to fuse the learning methods of both for image recognition research. This paper mainly introduces the current situation of zero-sample and small-sample learning and summarizes the learning of zero-sample and small-sample. And the meaning of zero-sample learning and small-sample learning and the classification of the main learning methods are introduced and compared and outlined, respectively. Finally, the methods of zero-sample and small-sample learning are fused, the design is introduced and analyzed, and the future research directions are prospected according to the current research problems.

Self-supervised patient-specific features learning for OCT image classification.

Deep learning's great success in image classification is heavily reliant on large-scale annotated datasets. However, obtaining labels for optical coherence tomography (OCT) data requires the significant effort of professional ophthalmologists, which hinders the application of deep learning in OCT image classification. In this paper, we propose a self-supervised patient-specific features learning (SSPSF) method to reduce the amount of data required for well OCT image classification results. Specifically, the SSPSF consists of a self-supervised learning phase and a downstream OCT image classification learning phase. The self-supervised learning phase contains two self-supervised patient-specific features learning tasks. One is to learn to discriminate an OCT scan which belongs to a specific patient. The other task is to learn the invariant features related to patients. In addition, our proposed self-supervised learning model can learn inherent representations from the OCT images without any manual labels, which provides well initialization parameters for the downstream OCT image classification model. The proposed SSPSF achieves classification accuracy of 97.74% and 98.94% on the public RETOUCH dataset and AI Challenger dataset, respectively. The experimental results on two public OCT datasets show the effectiveness of the proposed method compared with other well-known OCT image classification methods with less annotated data.

Numerical Analysis of Variations Distance Formulas on K Nearest Neighbors In Classifying Malaria Parasite Blood Cells

Malaria is one of the numerous acute and chronic diseases. Even malaria can pose a threat to a person's safety. The original cause of malaria was an infection with a protozoan of the genus Plasmodium, which was transmitted by the bite of a mosquito. This Anopheles mosquito parasite infects red blood cells throughout the body, resulting in an enlarged spleen. This research aims to make it easier for physicians to classify blood images as malaria-infected or not. If the input is a blood image, then SURF Feature Extraction will be used to extract the blood image. We therefore obtained weight results based on the extraction results. The weighted results generated by the SURF extraction process will be classified using the KNN Algorithm to determine whether or not an individual is infected with malaria. This study's tests compared various distance formulas utilized by the KNN classification method. Comparing the results of malaria blood image classification using the KNN classification method with variations in the distance formula, it is evident in table 7 that correlation is the optimal distance formula for malaria parasite blood cells recognition, followed by cosine. According to the results of KNN's tests, it is not optimal at classifying blood images containing malaria, but these results are categorized as good

Analysis of deforestation as impact of changes on oil palm land use in Tanah Bumbu Regency, South Kalimantan using satellite remote sensing data

Analysis of deforestation using remote sensing satellite imagery is carried out for monitoring and as an effort to reduce deforestation rates to maintain the ecological function of the forest. One of the causes of deforestation are land clearing and land conversion. Based on BPS data in 2020th, the forest area is decreasing and the area of oil palm is increasing every year. This research was conducted in Tanah Bumbu Regency, the data used were Landsat 7 ETM+ and Landsat 8 OLI/TIRS images. The method used is a guided classification with Support Vector Machine (SVM). Analysis of deforestation uses the results of image processing classification in time series by taking into account the reduction of forest that occurs every year. Based on the results of image classification in 2010th using SVM, forest has an area of 2161.1 km2 and oil palm is 1063.6 km2. Both of these land uses underwent changes in 2020th with forest area experiencing a reduction of 1493.42 km2 and oil palm having an additional area of 865.62 km2. The classification results have an overall accuracy value of above 80% and a kappa accuracy of 78%. The results of the analysis show that there is a relationship between the incidence of deforestation and the expansion of oil palm land. Based on the results of the regression analysis, R2 = 0.95 with a change value of 935.42 km2.

Obtaining Calibrated Probabilities with Personalized Ranking Models

For personalized ranking models, the well-calibrated probability of an item being preferred by a user has great practical value. While existing work shows promising results in image classification, probability calibration has not been much explored for personalized ranking. In this paper, we aim to estimate the calibrated probability of how likely a user will prefer an item. We investigate various parametric distributions and propose two parametric calibration methods, namely Gaussian calibration and Gamma calibration. Each proposed method can be seen as a post-processing function that maps the ranking scores of pre-trained models to well-calibrated preference probabilities, without affecting the recommendation performance. We also design the unbiased empirical risk minimization framework that guides the calibration methods to learning of true preference probability from the biased user-item interaction dataset. Extensive evaluations with various personalized ranking models on real-world datasets show that both the proposed calibration methods and the unbiased empirical risk minimization significantly improve the calibration performance.

Fast Convolutional Neural Network with iterative and non-iterative learning

Convolutional Neural Networks (CNNs) have achieved potentially good results for image classification. Due to their learning capabilities such networks are explored and implemented in real world applications. However, when it comes to training a CNN on large datasets without transfer learning, it takes long time and resources, due to the iterative process of weight update in fully connected layer. The backpropagation algorithm used to train the entire network suffers from slow convergence, getting trapped in a local minimum, not guaranteed to find global minimum and being hypersensitive to the learning rate set-up. Therefore, in this paper we propose a novel architecture of convolutional neural network with a non-iterative radial basis function-based classification layer which makes it more efficient for image classification tasks. This is a partially parameter free direct learning approach, which is highly beneficial in real-world applications as well as learning associated with large datasets. The proposed approach has been evaluated on four benchmark datasets such as CIFAR-10, MNIST, Digit and CE-MRI. It has achieved the accuracy scores of 99.8%, 99.8%, 98.4% and 98.3% on MNIST, Digit, CE-MRI and CIFAR-10 datasets respectively, when ResNet18 was used as backbone. Experimental results have illustrated that the proposed architecture is much better as it requires less parameter tuning, does not require handcrafted features and achieves same or higher accuracy in relatively lesser time than the standard CNN. Statistical significance tests were carried out to prove the efficiency of the proposed approach.

Convolutional Neural Network Compression through Generalized Kronecker Product Decomposition

Modern Convolutional Neural Network (CNN) architectures, despite their superiority in solving various problems, are generally too large to be deployed on resource constrained edge devices. In this paper, we reduce memory usage and floating-point operations required by convolutional layers in CNNs. We compress these layers by generalizing the Kronecker Product Decomposition to apply to multidimensional tensors, leading to the Generalized Kronecker Product Decomposition (GKPD). Our approach yields a plug-and-play module that can be used as a drop-in replacement for any convolutional layer. Experimental results for image classification on CIFAR-10 and ImageNet datasets using ResNet, MobileNetv2 and SeNet architectures substantiate the effectiveness of our proposed approach. We find that GKPD outperforms state-of-the-art decomposition methods including Tensor-Train and Tensor-Ring as well as other relevant compression methods such as pruning and knowledge distillation.

A Deep Learning-Based Model for Date Fruit Classification

A total of 8.46 million tons of date fruit are produced annually around the world. The date fruit is considered a high-valued confectionery and fruit crop. The hot arid zones of Southwest Asia, North Africa, and the Middle East are the major producers of date fruit. The production of dates in 1961 was 1.8 million tons, which increased to 2.8 million tons in 1985. In 2001, the production of dates was recorded at 5.4 million tons, whereas recently it has reached 8.46 million tons. A common problem found in the industry is the absence of an autonomous system for the classification of date fruit, resulting in reliance on only the manual expertise, often involving hard work, expense, and bias. Recently, Machine Learning (ML) techniques have been employed in such areas of agriculture and fruit farming and have brought great convenience to human life. An automated system based on ML can carry out the fruit classification and sorting tasks that were previously handled by human experts. In various fields, CNNs (convolutional neural networks) have achieved impressive results in image classification. Considering the success of CNNs and transfer learning in other image classification problems, this research also employs a similar approach and proposes an efficient date classification model. In this research, a dataset of eight different classes of date fruit has been created to train the proposed model. Different preprocessing techniques have been applied in the proposed model, such as image augmentation, decayed learning rate, model checkpointing, and hybrid weight adjustment to increase the accuracy rate. The results show that the proposed model based on MobileNetV2 architecture has achieved 99% accuracy. The proposed model has also been compared with other existing models such as AlexNet, VGG16, InceptionV3, ResNet, and MobileNetV2. The results prove that the proposed model performs better than all other models in terms of accuracy.

Discrimination of Breast Cancer Based on Ultrasound Images and Convolutional Neural Network.

The aim of our study was to establish an artificial intelligence tool for the diagnosis of breast disease base on ultrasound (US) images. A deep learning algorithm Efficient-Det assisted US diagnosis method was developed to determine breast suspicious lesions as benign, malignant, or normal. Totally 1181 US images from 487 patients of our hospital and 694 publicly accessible images were employed for modeling, including 558 benign images, 370 malignant images, and 253 normal tissue images. The actual diagnosis results for the patients were determined by the biopsy or surgery. Efficient-Det was first retrained using an exclusive public breast cancer US dataset with transfer learning techniques. A blind test set consisting of 50 benign, 50 malignant, and 50 normal tissue images was randomly picked from the patients' images as the independent test set to test its searching ability on suspicious tumor regions. Furthermore, the confusion matrix and classification accuracy were employed as evaluation metrics to select the optimal classification models. Efficient-Det has demonstrated remarkable progress in general image recognition tasks with specific advantages of locating and identifying tumor areas simultaneously. Compared to the manual method (mean accuracy: 95.3% and 60 s per image) and traditional feature engineering method (mean accuracy: 90% and 15 s per image), our Efficient-Det has the capability of providing a competitive (mean accuracy: 92.6%) and fast (0.06 s per image) classification result. The deployment of Efficient-Det in our local breast cancer discrimination task exhibits specific applicability within real clinical workflows.

Monitoring of Soil Salinization in the Keriya Oasis Based on Deep Learning with PALSAR-2 and Landsat-8 Datasets

Currently, soil salinization is one of the main forms of land degradation and desertification. Soil salinization not only seriously restricts the development of agriculture and the economy, but also poses a threat to the ecological environment. The main purpose of this study is to map soil salinity in Keriya Oasis, northwestern China using the PALSAR-2 fully polarized synthetic aperture radar (PolSAR) L-band data and Landsat8-OLI (OLI) optical data combined with deep learning (DL) methods. A field survey is conducted, and soil samples are collected from 20 April 2015 to 1 May 2015. To mine the hidden information in the PALSAR-2 data, multiple polarimetric decomposition methods are implemented, and a wide range of polarimetric parameters and synthetic aperture radar discriminators are derived. The radar vegetation index (RVI) is calculated using PALSAR-2 data, while the normalized difference vegetation index (NDVI) and salinity index (SI) are calculated using OLI data. The random forest (RF)-integrated learning algorithm is used to select the optimal feature subset composed of eight polarimetric elements. The RF, support vector machine, and DL methods are used to extract different degrees of salinized soil. The results show that the OLI+PALSAR-2 image classification result of the DL classification was relatively good, having the highest overall accuracy of 91.86% and a kappa coefficient of 0.90. This method is helpful to understand and monitor the spatial distribution of soil salinity more effectively to achieve sustainable agricultural development and ecological stability.

Attention-based CNNs for Image Classification: A Survey

Deep learning techniques as well as CNNs can learn power context information, they have been widely applied in image recognition. However, deep CNNs may reply on large width and large depth, which may increase computational costs. Attention mechanism fused into CNNs can address this problem. In this paper, we summary an attention mechanism acts a CNN for image classification. Firstly, the survey shows the development of CNNs for image classification. Then, we illustrate basis of CNNs and attention mechanisms for image classification. Next, we give the main architecture of CNNs with attentions, public and our collected datasets, experimental results in image classification. Finally, we point out potential research points, challenges attention-based for image classification and summary the whole paper.

A SAR-GMTI Approach Aided by Online Knowledge With an Airborne Multichannel Quad-Pol Radar System

In the complicated geographical environment, there will be a seriously deleterious effect to the performance of synthetic aperture radar (SAR)-ground moving target indication (SAR-GMTI) system, because it is difficult to obtain the homogeneous training samples to accurately estimate the clutter covariance matrix (CCM) without prior information of the observed scene. To this end, this paper proposes a SAR-GMTI approach aided by online knowledge with an airborne multichannel quadrature-polarimetric (quad-pol) radar system. Generally, this paper can be divided into two parts: online knowledge acquisition and polarization knowledge-aided (Pol-KA) SAR-GMTI processing. Firstly, based on the similarity of pixels from the multichannel and multi-polarization information, a weighed estimation method of polarimetric coherency matrix is proposed, which can overcome the over-smoothing problem and increase the estimation accuracy of coherency matrix. Further, a hybrid weighted local K-means based on geodesic distance (GD-HWLKM) clustering algorithm is proposed to achieve the aim of unsupervised classification. Here, geodesic distance (GD) is exploited to measure the distance between multi-feature region covariance matrixes (MFRCMs) and a hybrid weight from different scales (including local cover class distribution, region and pixel) is calculated to automatically update the cluster centroid, which can make full use of the local spatial information by taking the interclass samples' similarity and the diversity of different classes into consideration. Secondly, with the assistance of the previous polarization SAR (PolSAR) image classification result, a Pol-KA SAR-GMTI method is developed. For each ground cover category, an accurate CCM can be estimated with the independent and identically distributed (IID) training samples. Then, the multichannel clutter suppression and preliminary constant false alarm rate (CFAR) detection are performed. Finally, with an airborne multichannel quad-pol radar system, the experimental results on real measured data demonstrate that the proposed method can efficiently improve the clutter suppression preformation and moving-target detection preformation.

PolSAR Image Classification With Multiscale Superpixel-Based Graph Convolutional Network

Convolutional neural networks (CNNs) have demonstrated impressive ability to achieve promising results in PolSAR image classification. However, the traditional CNN performs convolution on local square regions with fixed sizes. The selection of these local square regions (patches) cannot fully take advantage of the boundary information of land covers and cannot search optimal neighborhoods in the whole image. To overcome these shortcomings, we propose a superpixel-based graph convolutional network (SP-GCN) for PolSAR image classification. SP-GCN utilizes superpixels as graph nodes, which makes full use of boundary information of superpixels and significantly reduces the computational cost of GCN, making it possible to apply GCN to large-scale PolSAR image classification. To reduce the impact of superpixel scale on classification results, we further propose a multiscale superpixel-based graph convolutional network (MSSP-GCN) based on the SP-GCN. Experimental results on three PolSAR datasets firmly demonstrate the superiority of the proposed SP-GCN and MSSP-GCN to other state-of-the-art methods.

DRFL-VAT: Deep Representative Feature Learning With Virtual Adversarial Training for Semisupervised Classification of Hyperspectral Image

While deep learning algorithms have achieved good results in hyperspectral image (HSI) classification, several supervised classification algorithms rely on a large number of labeled samples to get adequate performance. Collecting a large number of labeled samples is expensive in many real applications. To address this issue, a novel semisupervised HSI classification framework called deep representative feature learning (DRFL) with virtual adversarial training (DRFL-VAT) is developed in this article. By embedding the local manifold learning (LML) into the fully connected layers of a convolutional neural network (CNN), our newly developed DRFL can learn representative features. The VAT regularization is adopted to exploit the prediction label distribution of training samples and addresses the overfitting problem. Finally, the objective function of DRFL-VAT is solved by a customized algorithm. We test our method on three widely public HSI datasets and our results show that our method is competitive when compared to other state-of-the-art approaches.

HyperViTGAN: Semisupervised Generative Adversarial Network With Transformer for Hyperspectral Image Classification

Generative adversarial networks (GANs) have achieved many excellent results in hyperspectral image (HSI) classification in recent years, as GANs can effectively solve the dilemma of limited training samples in HSI classification. However, due to the class imbalance problem of HSI data, GANs always associate minority-class samples with fake label. To address this issue, we first propose a semi-supervised generative adversarial network incorporating a transformer, called HyperViTGAN. The proposed HyperViTGAN is designed with an external semi-supervised classifier to avoid self-contradiction when the discriminator performs both classification and discrimination tasks. The generator and discriminator with skip connection are utilized to generate HSI patches by adversarial learning. The proposed HyperViTGAN captures semantic context and low-level textures to reduce the loss of critical information. In addition, the generalization ability of HyperViTGAN is improved through the use of data augmentation. Experimental results on three well-known HSI data sets, Houston 2013, Indian Pines 2010, and Xuzhou, show that the proposed model achieves competitive HSI classification performance in comparison with the current state-of-the-art classification models.

Low Complexity Encoder with Multilabel Classification and Image Captioning燤odel

Due to the advanced development in the multimedia-on-demand traffic in different forms of audio, video, and images, has extremely moved on the vision of the Internet of Things (IoT) from scalar to Internet of Multimedia Things (IoMT). Since Unmanned Aerial Vehicles (UAVs) generates a massive quantity of the multimedia data, it becomes a part of IoMT, which are commonly employed in diverse application areas, especially for capturing remote sensing (RS) images. At the same time, the interpretation of the captured RS image also plays a crucial issue, which can be addressed by the multi-label classification and Computational Linguistics based image captioning techniques. To achieve this, this paper presents an efficient low complexity encoding technique with multi-label classification and image captioning for UAV based RS images. The presented model primarily involves the low complexity encoder using the Neighborhood Correlation Sequence (NCS) with a burrows wheeler transform (BWT) technique called LCE-BWT for encoding the RS images captured by the UAV. The application of NCS greatly reduces the computation complexity and requires fewer resources for image transmission. Secondly, deep learning (DL) based shallow convolutional neural network for RS image classification (SCNN-RSIC) technique is presented to determine the multiple class labels of the RS image, shows the novelty of the work. Finally, the Computational Linguistics based Bidirectional Encoder Representations from Transformers (BERT) technique is applied for image captioning, to provide a proficient textual description of the RS image. The performance of the presented technique is tested using the UCM dataset. The simulation outcome implied that the presented model has obtained effective compression performance, reconstructed image quality, classification results, and image captioning outcome.