Few-shot segmentation aims to devise a generalizing model that segments query images from unseen classes during training with the guidance of a few support images whose class tally with the class of the query. There exist two domain-specific problems mentioned in the previous works, namely spatial inconsistency and bias towards seen classes. Taking the former problem into account, our method compares the support feature map with the query feature map at multi scales to become scale-agnostic. As a solution to the latter problem, a supervised model, called as base learner, is trained on available classes to accurately identify pixels belonging to seen classes. Hence, subsequent meta learner has a chance to discard areas belonging to seen classes with the help of an ensemble learning model that coordinates meta learner with the base learner. We simultaneously address these two vital problems for the first time and achieve state-of-the-art performances on both PASCAL-5iand COCO-20idatasets.

Real-world datasets commonly have noisy labels, which negatively affects the performance of deep neural networks (DNNs). In order to address this problem, we propose a label noise robust learning algorithm, in which the base classifier is trained on soft-labels that are produced according to a meta-objective. In each iteration, before conventional training, the meta-training loop updates soft-labels so that resulting gradients updates on the base classifier would yield minimum loss on meta-data. Soft-labels are generated from extracted features of data instances, and the mapping function is learned by a single layer perceptron (SLP) network, which is called MetaLabelNet. Following, base classifier is trained by using these generated soft-labels. These iterations are repeated for each batch of training data. Our algorithm uses a small amount of clean data as meta-data, which can be obtained effortlessly for many cases. We perform extensive experiments on benchmark datasets with both synthetic and real-world noises. Results show that our approach outperforms existing baselines. The source code of the proposed model is available at https://github.com/gorkemalgan/MetaLabelNet.

The existence of noisy labels in the dataset causes significant performance degradation for deep neural networks (DNNs). To address this problem, we propose a Meta Soft Label Generation algorithm called MSLG, which can jointly generate soft labels using meta-learning techniques and learn DNN parameters in an end-to-end fashion. Our approach adapts the meta-learning paradigm to estimate optimal label distribution by checking gradient directions on both noisy training data and noise-free meta-data. In order to iteratively update soft labels, meta-gradient descent step is performed on estimated labels, which would minimize the loss of noise-free meta samples. In each iteration, the base classifier is trained on estimated meta labels. MSLG is model-agnostic and can be added on top of any existing model at hand with ease. We performed extensive experiments on CIFAR10, Clothing1M and Food101N datasets. Results show that our approach outperforms other state-of-the-art methods by a large margin. Our code is available at https://github.com/gorkemalgan/MSLG_noisy_label.

We propose a nonlinear kernel version of recently introduced basic thresholding classifier (BTC) for hyperspectral image (HSI) classification. BTC is a sparsity-based linear classifier, which utilizes inner product for similarity measure based on the fact that the hyperspectral data are linearly separable in the feature space. In practice, the pixels of different classes of a given HSI are not always distinct and may overlap. This is known as the inseparability problem, which reduces the performance of a linear classifier. In order to solve this problem in the feature space, we propose the kernel BTC (KBTC) algorithm that achieves promising performance by utilizing well-known “kernel trick.” To increase the classification accuracy further, we present spatial-spectral KBTC that incorporates spatial information using weighted least squares filter with a guidance image. Furthermore, we apply L0smoothing technique on the guidance image, which provides additional improvements. Experimental results on publicly available HSI data sets showed that the proposal and its spatial extension yield better classification results as compared with linear similarity-based BTC, nonlinear kernel-based support vector machines, kernel orthogonal matching pursuit, recently proposed logistic regression via splitting and augmented Lagrangian, and their spatial extensions.

The authors present a sparsity‐based algorithm, basic thresholding classifier (BTC), for classification applications which is capable of identifying test samples extremely rapidly and performing high classification accuracy. They introduce a sufficient identification condition (SIC) under which BTC can identify any test sample in the range space of a given dictionary. By using SIC, they develop a procedure which provides a guidance for the selection of threshold parameter. By exploiting rapid classification capability, they propose a fusion scheme in which individual BTC classifiers are combined to produce better classification results especially when very small number of features is used. Finally, they propose an efficient validation technique to reject invalid test samples. Numerical results in face identification domain show that BTC is a tempting alternative to sparsity‐based classification algorithms such as greedy orthogonal matching pursuit and l 1 ‐minimisation.

We propose a lightweight sparsity-based algorithm, namely, the basic thresholding classifier (BTC), for hyperspectral image (HSI) classification. BTC is a pixelwise classifier which uses only the spectral features of a given test pixel. It performs the classification using a predetermined dictionary consisting of labeled training pixels. It then produces the class label and residual vector of the test pixel. Since incorporating spatial and spectral information in HSI classification is quite an effective way of improving classification accuracy, we extend our proposal to a three-step spatial-spectral framework. First, every pixel of a given HSI is classified using BTC. The resulting residual vectors form a cube which could be interpreted as a stack of images representing residual maps. Second, each residual map is filtered using an averaging filter. Finally, the class label of each test pixel is determined based on minimal residual. Numerical results on public data sets show that our proposal outperforms well-known support vector machine-based techniques and sparsity-based greedy approaches like simultaneous orthogonal matching pursuit in terms of both classification accuracy and computational cost.

In this paper, a real-time railway fastener detection system using a high-speed laser range finder camera is presented. First, an extensive analysis of various methods based on pixel-wise and histogram similarities are conducted on a specific railway route. Then, a fusing stage is introduced which combines least correlated approaches also considering the performance upgrade after fusing. Then, the resulting method is tested on a larger database collected from a different railway route. After observing repeated successes, the method is implemented on NI LabVIEW and run real-time with a high-speed 3-D camera placed under a railway carriage designed for railway quality inspection.

An automatic segmentation algorithm that is based on watersheds and region merging type multi-scale segmentation (MSS) is proposed, which can be used as the initial step of an object-based classifier. First, the image is segmented using watershed segmentation. Then, primitive segments are merged to create meaningful objects by the proposed hybrid region merging algorithm. During these steps, an unsupervised segmentation accuracy metric is considered so that best performing parameters of the proposed algorithm are determined automatically. By this way, the proposed segmentation algorithm has become fully automatic. Experiments are done on images taken from Google Earth® software. Algorithm performance is computed using the ground truths of these images. Also, results for some other Google Earth® images are presented for qualitative performance assessment.

This paper proposes a new method for the classification of synthetic aperture radar (SAR) images based on a novel feature vector. The method aims at combining the intensity information of pixels with spatial information and structural relationships. Unlike classical approaches which define a static neighborhood via a rectangular moving window of predefined size and relate spatial information for each center pixel to all the pixels within that window, the local primitives (LPs) proposed in this study provide us with an adaptive neighborhood so that spatial information for each center pixel is extracted only from the related pixels in its neighborhood. LPs correspond to local homogeneous connected components that describe the pixel neighborhood more consistently than the fixed size window approach. A feature vector, called as the LP pattern (LPP), is constructed for each pixel. The feature vector includes information about the sizes, intensity levels, and contrast differences of LPs within a disk whose center is the pixel under consideration as well as the repetitive frequency of LPs outside that disk. Finally, a kernel-based support vector machine is used with the proposed feature vectors for the classification of SAR images. Experimental analysis presents that the new feature extraction technique is well suited to depict spatial information and structural relationships and it yields promising results for the classification of SAR images when compared to common features such as gray-level co-occurrence matrix and Gabor coefficients.

This paper proposes a new method for the classification of synthetic aperture radar (SAR) images based on a novel feature vector. The method aims at combining intensity information of pixels with spatial information and structural relationships. Unlike classical approaches which define a static neighborhood and relate spatial information for each center pixel to all the pixels within that window, the local primitives (LPs) proposed in this study provide us with an adaptive neighborhood for each pixel. LPs correspond to a certain number of layers of local homogenous connected components. Using LPs, a feature vector (local primitive pattern, LPP) is constructed for each pixel. The feature vector includes information about the sizes and contrast differences of LPs within a disk as well as the repetitive frequency of LPs outside that disk. To test the efficiency of LPP, support vector machine (SVM) classification is utilized.

This paper presents a classification method for hyperspectral images utilizing Differential Morphological Profiles (DMPs) which permit to include in the analysis spatial information since they can provide an estimate of the size and contrast characteristics of the structures in an image. Due to the wide variety of objects present in a scene, the pixels belonging to the same semantic structure may not have homogeneous spatial and spectral features. In addition, instead of a single peak (which can be related to a measure of the scale), multiple local maxima and multiple responses are usually observed in the DMP. In order to handle such intra-class variability, class-specific weighting functions are employed in order to differently modulate the DMP values according to the different characteristics of the land cover types. In such way, it is possible to differentiate the behaviors of the DMP for each pixel in the image according to its semantic, providing an increase of the separability of the classes. At first, a DMP computed with opening by reconstruction (DMPO) and one with closing by reconstruction (DMPC) are derived on each of the first principle components extracted from the hyperspectral image. Then, both profiles are weighted by each class-specific weighting function and concatenated in a single data structure. The constructed feature vectors are considered by a random forest classifier.

This paper presents a segmentation method that exploits object based region merging to delineate the context of the image under investigation. There are two main steps. Initially, primitive objects are obtained by morphological operations that generate spectrally homogenous primitives. We assume that primitives are components of semantic objects that are of interest. Next, these primitives therefore are modeled and merged based on expectation maximization. We presented the results of the experiments applied to QuickBird images of rural and urban areas taken from the city of Ankara, Turkey. Experimental results demonstrate the capabilities of these methods along with their limitations.

ABSTRACT Biomechanical modeling of soft tissue is a complex problem for achieving realistic surgical simulations, surgical planning, and scientific analysis. In the literature, three categories of biomechanical models: spline based models, spring models, and finite element models (FEMs) are mainly used for dealing with this problem. Among these, spline based models offer relatively fast and realistic soft tissue simulations by utilizing both the spring and FEMs. In this paper, a new dynamic volume spline model for human face skin is proposed and the performance of our model is discussed by estimating the results of facial surgery of three different patients. Face models of the patients are obtained from 3D CT/MR scans by segmenting the skull, muscle, and skin layers. In these face models, the skull and the muscle layers are considered as the rigid boundary for the skin layer and the skin layer is modeled by our dynamic volume spline. The control points of the dynamic volume spline are localized masses with viscoelastic material properties (stiffness, damping, and mass). These parameters are computed from the skin material properties that were published in the literature. Once the face models are generated, facial surgery plannings are simulated. Infact, the pre‐surgery face models are modified according to the surgical plans and the estimated post‐surgery face models are compared with the actual post‐surgery face models. Moreover, in order to discuss the performance of our dynamic volume spline model, the same analyses are performed on the post‐surgery estimations of a conventional tool. Copyright © 2011 John Wiley & Sons, Ltd.

Using mean curvature (H) and Gaussian curvature (K) values or shape index (S) and curvedness (C) values, HK and SC curvature spaces are constructed in order to classify surface patches into types such as pits, peaks, saddles etc. Since both HK and SC curvature spaces classify surface patches in to similar types, their classification capabilities are comparable. Previously, HK and SC curvature spaces were compared in terms of their classification ability only at the given data resolution [2]. When calculating H, K, C and S values, the scale/resolution ratio is highly effective. However, due to its scale invariant nature, shape index (S) values are independent of the resolution or the scale. Thus it is no wonder that SC method gives better results than HK method when the comparison is carried out at an uncontrolled scale/resolution level. In this study, the scale/resolution ratio is set to a constant value for the whole database and scale spaces based on both HK and SC methods are built. Scale and orientation invariant features are extracted using scale spaces and these features are used in object recognition tasks. The methods are compared both mathematically and experimentally in terms of their surface classification and object recognition performances.

Finding slope units for a given watershed is an important task before analyzing landslide susceptibility. Usually, slope unit generation requires integration of several digital elevation model (DEM)-based outputs obtained from GIS and related hydrological software. Therefore, it is time consuming due to involved steps and compilation of various software outputs. In this paper a DEM-based, scale and resolution invariant, fully automatic method is proposed for creating slope units to be used in landslide susceptibility mapping.

An algorithm is proposed to extract transformation and scale invariant 3D fundamental elements from the surface structure of 3D range scan data. The surface is described by mean and Gaussian curvature values at every data point at various scales and a scale-space search is performed in order to extract the fundamental structures and to estimate the location and the scale of each fundamental structure. The extracted fundamental structures can later be used as nodes in a topological graph where the links between the nodes can be defined as the spatial and geometric relations between the fundamental elements.

An algorithm is proposed for 3D object representation using generic 3D features which are transformation and scale invariant. Descriptive 3D features and their relations are used to construct a graphical model for the object which is later trained and then used for detection purposes. Descriptive 3D features are the fundamental structures which are extracted from the surface of the 3D scanner output. This surface is described by mean and Gaussian curvature values at every data point at various scales and a scale-space search is performed in order to extract the fundamental structures and to estimate the location and the scale of each fundamental structure.

Many approaches to object recognition are founded on probability theory, and can be broadly characterized as either generative or discriminative according to whether or not the distribution of the image features is modelled. Generative and discriminative methods have very different characteristics, as well as complementary strengths and weaknesses. In this paper we introduce new generative and discriminative models for object detection and classification based on weakly labelled training data. We use these models to illustrate the relative merits of the two approaches in the context of a data set of widely varying images of non-rigid objects (animals). Our results support the assertion that neither approach alone will be sufficient for large scale object recognition, and we discuss techniques for combining them.