Erasing-inpainting-based data augmentation using denoising diffusion probabilistic models with limited samples for generalized surface defect inspection

The lack of training samples remains one of the major obstacles in applying convolutional neural networks (CNNs) to the hyperspectral image (HSI) classification. In this letter, the accurate classification of HSI with limited training samples is investigated. Due to the advantages of minimizing the distance between samples in the same class and maximizing the distance between samples in different classes, siamese CNN is used for HSI classification with limited training samples. After that, to improve the classification performance, data augmentation is investigated for siamese CNN-based HSI classification. Specifically, pair data augmentation based on CutMix is proposed to generate the training pairs of the same or different classes and the new generated training pairs are used to train siamese CNN. Traditional data augmentation methods simply generate new training samples. It is not proper when data augmentation methods keep the loss function and change the content of the input at the same time. Therefore, a soft-loss-based siamese CNN, which changes its loss according to the coupled replacement data augmentation, is proposed to further address the HSI classification with limited training samples. In the experimental part, on two widely used hyperspectral datasets, the influences of different training samples and window sizes are discussed, and the experimental results reveal that the proposed soft-augmentation-based siamese CNN provides competitive results with limited training samples compared with state-of-the-art methods.

Recently, Convolutional Neural Network (CNN) has achieved some success in synthetic aperture radar (SAR) image classification. This outstanding performance mainly depends on a large number of training samples, but achieved satisfactory classification results with limited training samples remains a challenge. To address this problem, we propose a dual-scale Siamese densely connected network with Markov Random Fields (DS-SDCNet-MRF) for single-polarization SAR image classification. First, the Siamese densely connected network (SDCNet) is proposed to fully extract discriminative features under limited samples. Then, the proposed DS-SDCNet is constructed with two SDCNets of different scales to produce complementary classification results. Among them, the large-scale SDCNet has a better classification result in the homogeneous region, while the small-scale SDCNet tends to provide good detail preservation. Finally, an improved MRF model which combined the category probability information is proposed to further improve the classification performance. Experimental results on simulated and real single-polarization SAR data demonstrate that the proposed method achieves more encouraging classification performance than the current state-of-the-art classification methods with limited samples.

Data augmentation (DA) increases the diversity of training data to improve the model generalization ability. Most of the DA methods focus on image classification or object detection. Directly using the existing DA methods on change detection tasks not only falls short of fully exploring the specificity of the change image pairs but also leads to longer training times. In this article, we first propose a mask-guided mixing (MGM) DA for change detection, which mixes the change regions of the current training sample based on prediction results and labels to generate high-quality samples with more positive samples. We then propose a new reinforcement learning (RL)-based Adaptive DA method, AdaAug+, to adaptively select the optimal DA policy for the training samples. An actor selects the best augmentation operation from the operation set according to the image pair. The augmented image pairs make it easier for the change detector to learn the optimal parameters and improve the final detection performance. To reduce the training time, we identify and remove the redundant training samples during the training process by our redundancy searching policy. We have conducted various experiments on four remote sensing change detection datasets with different change detectors. The experimental results demonstrate that AdaAug+ achieves promising performance compared to the state-of-the-art DA methods and requires less training time.

Availability of limited training remote sensing datasets is one of the problems in deep learning, as deep architectures require a large number of training samples for proper training. In this paper, we present a technique for data augmentation based on a spectral indexed generative adversarial network to train deep convolutional neural networks. This technique uses the spectral characteristic of multispectral (MS) images to support data augmentation in order to generate realistic training samples with respect to each land-use and land-cover class. The impact of multispectral remote sensing data generated through the spectral indexed GAN are evaluated through classification experiments. Experimental results obtained on the classification of the Sentinel-2 Eurosatallband datasets show that data augmentation through spectral indexed GAN enhances the main accuracy metrics.

Convolutional Neural Networks have been widely used in remote sensing scene classification. Since this kind of model needs a large number of training samples containing data category information, a Generative Adversarial Network (GAN) is usually used to address the problem of lack of samples. However, GAN mainly generates scene data samples that do not contain category information. To address this problem, a novel supervised adversarial Lie Group feature learning network is proposed. In the case of limited data samples, the model can effectively generate data samples with category information. There are two main differences between our method and the traditional GAN. First, our model takes category information and data samples as the input of the model and optimizes the constraint of category information in the loss function, so that data samples containing category information can be generated. Secondly, the object scale sample generation strategy is introduced, which can generate data samples of different scales and ensure that the generated data samples contain richer feature information. After large-scale experiments on two publicly available and challenging datasets, it is found that our method can achieve better scene classification accuracy even with limited data samples.

Traditional hyperspectral image (HSI) classification methods mainly include machine learning and convolutional neural network. However, they extremely depend on the large training samples. To obtain high accuracy on limited training samples, we propose a novel end-to-end spectral-spatial multi-scale network (SSMSN) for HSI classification. The SSMSN uses the multi-scale spectral module and the multi-scale spatial module to extract discriminative multi-scale spectral and multi-scale spatial features separately. In the multi-scale spectral module and spatial module, the multi-scale Res2Net block structure can learn multi-scale features at a granular level and increase the range of receptive fields by constructing the hierarchical residual-like connection within one single residual block. To alleviate the overfitting problem and further improve the classification accuracy on limited training samples, we adopt a simple but effective hinge cross-entropy loss function to train the SSMSN at the dynamic learning rate. A large number of experimental results demonstrate that on the Indiana Pines, University of Pavia, Kennedy Space Center, and Salinas Scene data sets, the proposed SSMSN achieves higher classification accuracy than state-of-the-art methods on limited training samples. Meanwhile, our SSMSN obtains less training and testing time than the popular AUSSC method.

Deep neural networks, albeit their great success on feature learning in various computer vision tasks, are usually considered as impractical for online visual tracking, because they require very long training time and a large number of training samples. In this paper, we present an efficient and very robust tracking algorithm using a single convolutional neural network (CNN) for learning effective feature representations of the target object in a purely online manner. Our contributions are multifold. First, we introduce a novel truncated structural loss function that maintains as many training samples as possible and reduces the risk of tracking error accumulation. Second, we enhance the ordinary stochastic gradient descent approach in CNN training with a robust sample selection mechanism. The sampling mechanism randomly generates positive and negative samples from different temporal distributions, which are generated by taking the temporal relations and label noise into account. Finally, a lazy yet effective updating scheme is designed for CNN training. Equipped with this novel updating algorithm, the CNN model is robust to some long-existing difficulties in visual tracking, such as occlusion or incorrect detections, without loss of the effective adaption for significant appearance changes. In the experiment, our CNN tracker outperforms all compared state-of-the-art methods on two recently proposed benchmarks, which in total involve over 60 video sequences. The remarkable performance improvement over the existing trackers illustrates the superiority of the feature representations, which are learned purely online via the proposed deep learning framework.

A problem with convolutional neural networks (CNNs) is that they require large datasets to obtain adequate robustness; on small datasets, they are prone to overfitting. Many methods have been proposed to overcome this shortcoming with CNNs. In cases where additional samples cannot easily be collected, a common approach is to generate more data points from existing data using an augmentation technique. In image classification, many augmentation approaches utilize simple image manipulation algorithms. In this work, we propose some new methods for data augmentation based on several image transformations: the Fourier transform (FT), the Radon transform (RT), and the discrete cosine transform (DCT). These and other data augmentation methods are considered in order to quantify their effectiveness in creating ensembles of neural networks. The novelty of this research is to consider different strategies for data augmentation to generate training sets from which to train several classifiers which are combined into an ensemble. Specifically, the idea is to create an ensemble based on a kind of bagging of the training set, where each model is trained on a different training set obtained by augmenting the original training set with different approaches. We build ensembles on the data level by adding images generated by combining fourteen augmentation approaches, with three based on FT, RT, and DCT, proposed here for the first time. Pretrained ResNet50 networks are finetuned on training sets that include images derived from each augmentation method. These networks and several fusions are evaluated and compared across eleven benchmarks. Results show that building ensembles on the data level by combining different data augmentation methods produce classifiers that not only compete competitively against the state-of-the-art but often surpass the best approaches reported in the literature.

Convolutional neural networks (CNNs) have been successfully applied to the single target tracking task in recent years. Generally, training a deep CNN model requires numerous labeled training samples, and the number and quality of these samples directly affect the representational capability of the trained model. However, this approach is restrictive in practice, because manually labeling such a large number of training samples is time-consuming and prohibitively expensive. In this article, we propose an active learning method for deep visual tracking, which selects and annotates the unlabeled samples to train the deep CNN model. Under the guidance of active learning, the tracker based on the trained deep CNN model can achieve competitive tracking performance while reducing the labeling cost. More specifically, to ensure the diversity of selected samples, we propose an active learning method based on multiframe collaboration to select those training samples that should be and need to be annotated. Meanwhile, considering the representativeness of these selected samples, we adopt a nearest-neighbor discrimination method based on the average nearest-neighbor distance to screen isolated samples and low-quality samples. Therefore, the training samples' subset selected based on our method requires only a given budget to maintain the diversity and representativeness of the entire sample set. Furthermore, we adopt a Tversky loss to improve the bounding box estimation of our tracker, which can ensure that the tracker achieves more accurate target states. Extensive experimental results confirm that our active-learning-based tracker (ALT) achieves competitive tracking accuracy and speed compared with state-of-the-art trackers on the seven most challenging evaluation benchmarks. Project website: https://sites.google.com/view/altrack/.

(Aim) Breast cancer is the most common cancer in women and the second most common cancer worldwide. With the rapid advancement of deep learning, the early stages of breast cancer development can be accurately detected by radiologists with the help of artificial intelligence systems. (Method) Based on mammographic imaging, a mainstream clinical breast screening technique, we present a diagnostic system for accurate classification of breast abnormalities based on ResNet-50. To improve the proposed model, we created a new data augmentation framework called SCDA (Scaling and Contrast limited adaptive histogram equalization Data Augmentation). In its procedure, we first conduct the scaling operation to the original training set, followed by applying contrast limited adaptive histogram equalisation (CLAHE) to the scaled training set. By stacking the training set after SCDA with the original training set, we formed a new training set. The network trained by the augmented training set, was coined as ResNet-SCDA-50. Our system, which aims at a binary classification on mammographic images acquired from INbreast and MINI-MIAS, classifies masses, microcalcification as "abnormal", while normal regions are classified as "normal". (Results) We present the first attempt to use the image contrast enhancement method as the data augmentation method, resulting in an averaged 98.55 percent specificity and 92.83 percent sensitivity, which gives our best model an overall accuracy of 95.74 percent. (Conclusion) Our proposed method is effective in classifying breast abnormality.

Neural networks (NNs) are efficient tools for rapidly obtaining geoelectric models to solve magnetotelluric (MT) inversion problems. Training an NN with strong predictive power requires numerous training samples to prevent underfitting. To reduce the computational burden of generating a large number of training samples, this work analyzes the influence of the sample features and distribution on the training effect for an NN and proposes an efficient method of sample generation. This innovative method consists of three steps: 1) geoelectrically simplifying the features; 2) removing unnecessary features on the basis of realistic geological characteristics; and 3) mapping the samples to a higher-dimensional space. Numerical examples based on simple stratified models show that the number of samples can be reduced to below one-millionth of the original number while improving the predictive effect of the NN. The performance and effectiveness for processing more complex structures are verified by the inversion results obtained for a public data set, COPROD2. We conclude that this advanced method can generate high-quality training samples at a greatly reduced computational cost. The analysis of the sample features and distribution not only advances the state of research on the use of machine learning in geophysical inversion but also is a forward-looking study on the mechanisms of underfitting, tracing the source of these phenomena back to the training samples used.

Learning-enabled spectroscopic analysis, promising for automated real-time analysis of chemicals, is facing several challenges. First, a typical machine learning model requires a large number of training samples that physical systems can not provide. Second, it requires the testing samples to be in range with the training samples, which often is not the case in the real world. Further, a spectroscopy device is limited by its memory size, computing power, and battery capacity. That requires highly efficient learning models for on-site analysis. In this paper, by analyzing multi-gas mixtures and multi-molecule suspensions, we first show that orders of magnitude reduction of data dimension can be achieved as the number of principal components that need to be retained is the same as the independent constituents in the mixture. From this principle, we designed highly compact models in which the essential principal components can be directly extracted from the interrelations between the individual chemical properties and principal components; and only a few training samples are required. Our model can predict the constituent concentrations that have not been seen in the training dataset and provide estimations of measurement noises. This approach can be extended as an effectively standardized method for principle component extraction.

Based on margin sampling (MS) strategy, an active learning approach was introduced for proposed sample selection from large quantities of labeled samples using a Landsat-7 ETM+ image to solve remote sensing image classification problems for large number of training samples. As a breakthrough from conventional random sampling and stratified systematic sampling methods, this approach ensures classification of only using a few hundred training samples to be as effective as that of using several thousand and even tens of thousands of samples by conventional methods, thereby avoiding enormous calculations, substantially reducing operating time and improving training efficiency. The test results of the proposed approach was compared with those of random sampling and stratified systematic sampling, and the effects of training samples on classification under optimized and non-optimized selection conditions was analyzed.

Patch-level contrastive embedding learning for respiratory sound classification

Data augmentation is widely utilized as an effective technique to enhance the generalization performance of deep models. However, data augmentation may inevitably introduce distribution shifts and noises, which significantly constrain the potential and deteriorate the performance of deep networks. To this end, we propose a novel information-preserving framework, namely IPF-RDA, to enhance the robustness of data augmentations in this paper. IPF-RDA combines the proposal of (i) a new class-discriminative information estimation algorithm that identifies the points most vulnerable to data augmentation operations and corresponding importance scores; And (ii) a new information-preserving scheme that preserves the critical information in the augmented samples and ensures the diversity of augmented data adaptively. We divide data augmentation methods into three categories according to the operation types and integrate these approaches into our framework accordingly. After being integrated into our framework, the robustness of data augmentation methods can be enhanced and their full potential can be unleashed. Extensive experiments demonstrate that although being simple, IPF-RDA consistently improves the performance of numerous commonly used state-of-the-art data augmentation methods with popular deep models on a variety of datasets, including CIFAR-10, CIFAR-100, Tiny-ImageNet, CUHK03, Market1501, Oxford Flower, and MNIST, where its performance and scalability are stressed.