Object Classification Problems Research Articles

Learning Offset Probability Distribution for Accurate Object Detection

Object detection combines object classification and object localization problems. Current object detection methods heavily depend on regression networks to locate objects, which are optimized with various regression loss functions to predict offsets between candidate boxes and objects. However, these regression losses are difficult to assign the appropriate penalties for samples with large offset errors, resulting in suboptimal regression networks and inaccurate object offsets. In this article, we consider object location as offset bin classification problem, and propose a distance-aware offset bin classification network optimized with multiple binary cross entropy losses to learn various offset probability distribution, including single label distribution and distance-aware label distribution. On one hand, it provides gradient contributions for different samples based on the bounded probability instead of previous incalculable offset error. On the other hand, it explores the distance correlations between discrete offset bins to facilitate network learning. Specifically, we discretize the continuous offset into a number of bins, and predict the probability of each offset bin, in which the probability should be higher for the offset bin closer to the target offsets, and vice versa. Furthermore, we propose an expectation-based offset prediction and a hierarchical focusing method to improve the precision of prediction. We conduct extensive experiments to evaluate the effectiveness of our method. In addition, our method can be conveniently and flexibly inserted into existing object detection methods, which consistently achieves a large gain based on popular anchor-based and anchor-free methods on the PASCAL VOC, MS-COCO, KITTI, and CrowdHuman datasets. Code will be released at: https://github.com/QiuHeqian/DBC .

Transfer Learning and Few-Shot Learning Based Deep Neural Network Models for Underwater Sonar Image Classification With a Few Samples

Acoustic imaging sonar systems are widely used for long-range underwater surveillance in various civilian and military applications. They provide 2-D images of underwater objects, even in turbid water conditions where optical underwater imaging systems fail. Achieving high accuracy in automatic deep learning based underwater image classification remains an open problem due to insufficient data availability, poor image resolution, low signal-to-noise ratio surroundings, etc. In this study, we conduct a comparative analysis of different advanced deep learning approaches, i.e., transfer learning and few-shot learning, to address the problem of automatic object classification in sonar images, using a few samples of data. Specifically, two metric learning-based approaches, i.e., siamese network and triplet network as well as library-based approaches, are studied under the few-shot learning paradigm. Extensive experiments are conducted on a novel custom-made dataset developed in-house, along with the publicly available SeabedObjectsKLSG dataset. In addition, the effectiveness of the sampling technique in handling class imbalance during model training is also investigated in this work. Our experimental results highlight that the few-shot learning based approach is a promising direction for future research on underwater image classification with a few samples.

Enhancing functional efficiency in information-extreme machine learning with logistic regression ensembles

The subject matter of the article is the application of supervised machine learning for the task of object class recognition. The goal is enhancing functional efficiency in information-extreme technology (IET) for object class recognition. The tasks to be solved are: to analyse possible ways of increasing the functional efficiency IET approach; implement an ensemble of models that include logistic regression for prioritizing recognition features and an IEI learning algorithm; compare the functional efficiency of the resulting ensemble of models on well-known dataset with classic approach and results of other researchers. The methods: The method is developed within the framework of the functional approach to modelling natural intelligence applied to the problem of object classification. The following results were obtained: The study tries to augment existing IET to support feature prioritization as part of the object class recognition algorithm. The classical information-extreme algorithm treats all input features equivalently important in forming the decisive rule. As a result, the object features with strong correlation are not prioritized by the algorithm's decisive mechanism – resulting in decreasing functional efficiency in exam mode. The proposed approach is solving this problem by applying a two-stage approach. In the first stage the multiclass logistic regression applied to the input training features vectors of objects to be classified – formed the normalized training matrix. To prevent overfitting of the logistic regression, a model the L2(ridge) regularization method was used. On the second stage, the information-extreme method as input takes the result of the first stage. The geometrical parameters of class containers and the control tolerances on the recognition features were considered as the optimization parameters. Conclusions. The proposed approach increases MNIST (Modified National Institute of Standards and Technology) dataset classification accuracy compared with the classic information-extreme method by 26,44%. The proposed approach has a 3.77% lower accuracy compared to neural-like approaches but uses fewer resources in the training phase and allows retraining the model, as well as expanding the dictionary of recognition classes without model retraining.

Comparison of estimating vegetation index for outdoor free-range pig production using convolutional neural networks.

This study aims to predict the change in corn share according to the grazing of 20 gestational sows in a mature corn field by taking images with a camera-equipped unmanned air vehicle (UAV). Deep learning based on convolutional neural networks (CNNs) has been verified for its performance in various areas. It has also demonstrated high recognition accuracy and detection time in agricultural applications such as pest and disease diagnosis and prediction. A large amount of data is required to train CNNs effectively. Still, since UAVs capture only a limited number of images, we propose a data augmentation method that can effectively increase data. And most occupancy prediction predicts occupancy by designing a CNN-based object detector for an image and counting the number of recognized objects or calculating the number of pixels occupied by an object. These methods require complex occupancy rate calculations; the accuracy depends on whether the object features of interest are visible in the image. However, in this study, CNN is not approached as a corn object detection and classification problem but as a function approximation and regression problem so that the occupancy rate of corn objects in an image can be represented as the CNN output. The proposed method effectively estimates occupancy for a limited number of cornfield photos, shows excellent prediction accuracy, and confirms the potential and scalability of deep learning.

Bigisayarlı Görü Tabanlı AutoML Platformu

The rapid increase in data production, thanks to technological developments and scientific research, leads to the development of Machine Learning (ML) and similar new data analysis tools. It was announced that Amazon Web Services (AWS), a cloud service provider, stored 500EB of data in 2021 [1]. ML is an alternative to traditional engineering methods and does not require field knowledge of the problem to obtain a solution. However, the implementation of ML Algorithms can be complex depending on the content of the data set, and expert knowledge is the most important factor to use these algorithms effectively. Various methods have been developed to find a solution to this problem. There are many different areas and problems that machine learning can be applied to. We have limited our research to problems that can be solved using computer vision and AutoML. We have used AutoML and computer vision-based solutions to solve object classification, detection and segmentation problems. Our goal is to develop a platform that will work without the intervention of any expert. Users can load their datasets, choose the method they want, and train their models according to the problem they choose without any other intervention. After the training process is over, they can use their models in real time by transferring them over the platform in real time with their own hardware.

Mathematical Model of Multispectral Data Processing for a Mobile Ecology Monitoring Platform

The purpose of research is a mathematical justification of the process of processing multispectral data in order to detect local environmental pollution zones with the possibility of classifying the pollutant. Methods. The fundamentals of the applied theory of stochastic systems based on equations for multidimensional characteristic functions and functionals are used as a basic mathematical apparatus. When determining a contaminant, a criterion reflecting the ability of objects obeying Lambert's law is used. To solve the problem of object classification, approaches using binary logistic regression are applied. Statistical methods of analysis were used to evaluate the results of the study. Results. The obtained partial mathematical models allow us to take into account many factors affecting mobile environmental monitoring platforms operating in automatic mode. Substantiate the possibility of remote analysis of local environmental pollution zones, with the possibility of determining pollutants such as hydrocarbons, phosphate ions, etc., as well as searching for unauthorized locations of construction and household garbage. They increase the accuracy characteristics by 1,3 times when determining the parameters of selected objects due to the processing of data obtained in various spectral ranges. They contribute to reducing the computational complexity of the classification algorithm by 1,1 times, taking into account the volume of input data in a limited spectral range and reducing the resolution of the reference object, while not affecting the accuracy of classification. Conclusion. A mathematical model has been developed for processing data and images obtained in several spectral ranges during the operation of a multispectral device for an autonomous mobile environmental monitoring platform, which makes it possible to identify objects in the field of view of the device from a mobile platform, to obtain a detailed image of working scene objects with spatial reference relative to the coordinate system used, a distinctive feature of which is to increase the accuracy of calculating the coordinates of local zones pollution, and increasing the reliability of the classification of objects based on the characteristics of diffusive reflectivity in various spectral ranges.

Epistemic uncertainty aware semantic localization and mapping for inference and belief space planning

We investigate the problem of autonomous object classification and semantic SLAM, which generally exhibits a tight coupling between classification, metric SLAM, and planning under uncertainty. We contribute a unified framework for inference and belief space planning (BSP) that addresses prominent sources of uncertainty in this context: classification aliasing (classifier cannot distinguish between candidate classes from certain viewpoints), classifier epistemic uncertainty (classifier receives data “far” from its training set), and localization uncertainty (camera and object poses are uncertain). Specifically, two methods are developed for maintaining a joint distribution over robot and object poses, and over a posterior class probability vector that considers epistemic uncertainty in a Bayesian fashion. The first approach is Multi-Hybrid (MH), where multiple hybrid beliefs over poses and classes are maintained to approximate the joint belief over poses and posterior class probability. The second approach is Joint Lambda Pose (JLP), where the joint belief is maintained directly using a novel JLP factor. Furthermore, we extend both methods to BSP, while reasoning about future posterior epistemic uncertainty indirectly or directly via a novel information-theoretic reward function. Both inference methods utilize a novel viewpoint-dependent classifier uncertainty model that leverages the coupling between poses and classification scores and predicts the epistemic uncertainty from certain viewpoints. In addition, this model is used to generate predicted measurements during planning. To the best of our knowledge, this is the first work that reasons about classifier epistemic uncertainty within semantic SLAM and BSP. We extensively evaluate our inference and BSP approaches in simulation using real image data from the Active Vision Dataset. Results clearly indicate superior classification performance of our methods compared to an approach that is not epistemic uncertainty aware.

Ship Classification in Synthetic Aperture Radar Images Based on Multiple Classifiers Ensemble Learning and Automatic Identification System Data Transfer Learning

With the continuous development of earth observation technology, space-based synthetic aperture radar (SAR) has become an important source of information for maritime surveillance, and ship classification in SAR images has also become a hot research direction in the field of maritime ship monitoring. In recent years, the remote sensing community has proposed several solutions to the problem of ship object classification in SAR images. However, it is difficult to obtain an adequate amount of labeled SAR samples for training classifiers, which limits the application of machine learning, particularly deep learning methods, in SAR image ship object classification. In contrast, as a real-time automatic tracking system for monitoring ships at sea, a ship automatic identification system (AIS) can provide a large amount of relatively easy-to-obtain labeled ship samples. Therefore, to solve the problem of SAR image ship classification and improve the classification performance of learning models with limited samples, we proposed a SAR image ship classification method based on multiple classifiers ensemble learning (MCEL) and AIS data transfer learning. The core idea of our method is to transfer the MCEL model trained on AIS data to SAR image ship classification, which mainly includes three steps: first, we use the acquired global space-based AIS data to build a dataset for ship object classification models training; then, the ensemble learning model is constructed by combining multiple base classifiers; and finally, the trained classification model is transferred to SAR images for ship type prediction. Experiments show that the proposed method achieves a classification accuracy of 85.00% for the SAR ship classification, which is better than the performance of each base classifier. This proves that AIS data transfer learning can effectively solve the problem of SAR ship classification with limited samples, and has important application value in maritime surveillance.

Information-Theoretic Approach to the Problem of Automatic Image Recognition

The problem of automatic half-tone image recognition from the large database is formulated. Its decision using data on a problem of statistical classification of discrete objects on the basis of the minimum information discrimination principle is proposed. The new probabilistic model of a half-tone image is studied. On the basis of this model, a new image recognition algorithm using the minimum Kullback–Leibler information discrimination criterion is synthesized. It is proved that this criterion is optimal in Bayesian terms for the classification problem of discrete random objects, including digital images. The program and results of experimental research on the face recognition problem are presented. It is shown that the accuracy of image recognition with various light conditions on the basis of the offered criterion is essentially higher than the accuracy of traditional methods.

Brain tumor detection in MRI scans using single shot multibox detector

Researchers used visual methods rigorously to improve brain tumor detection in MRI or CT scans, yet there remains a challenge to improve the detection accuracy. Further, the rise of deep learning methods improved tumor detection accuracy up to the mark. But again, many times, we face the challenges of having a bigger dataset and better computing power to achieve an improved and accurate trained model for every object classification problem. In this paper, we propose a deep learning framework single shot multi-box detector (SSD)-based model to detect tumors in the MRI scans. The proposed SSD model is the faster algorithm to detect the tumor even with the ability to detect the smallest spot in the low-resolution MRI scans. We additionally used a lightweight neural network architecture MobileNet v2 with SSD for faster and accurate object classification. The experimental results showed 98% accuracy with the proposed method after training with the smallest dataset of 250 MRI scans. We used the Kaggle database for training and testing the proposed model.

MIYOLO: Modification of Improved YOLO-v3

In the world of detection and classification, You Only Look Once (YOLO)-v3 has been maintaining its position as the best classifier and detector and is due to its fastness, accuracy, less complex structure, continuous modifications, improvements, etc. Several approaches have been proposed in this paper to modify improved YOLO-v3. Our main contributions are (1) increasing the number of anchor boxes; (2) replacement of K-means clustering with density-based spatial clustering of applications with noise (DBSCAN); (3) introducing final bounding box alignment (FBBA) technique; and (4) replacement of multi-scaling with a depth calculation algorithm. At first, we increased the number of anchor boxes from 9 to 11 which resulted in a higher performance of 6% for both visual object classes and common objects in context datasets, where increment in computational cost was negligible. Similarly, it has been analyzed that a true number of anchor boxes can increase efficiency. So, instead of generating a static number of anchor boxes with k-means clustering, DBSCAN has been proposed for generating a dynamic and accurate number of anchor boxes. For minimizing the localization error, a new FBBA method has been introduced which would exactly overlap the object. In the same way, depth calculation algorithm can solve the small and large size object detection and classification problem without doing multi-scaling.

Evaluation of Automated Measurement of Hair Density Using Deep Neural Networks.

Recently, deep learning has been employed in medical image analysis for several clinical imaging methods, such as X-ray, computed tomography, magnetic resonance imaging, and pathological tissue imaging, and excellent performance has been reported. With the development of these methods, deep learning technologies have rapidly evolved in the healthcare industry related to hair loss. Hair density measurement (HDM) is a process used for detecting the severity of hair loss by counting the number of hairs present in the occipital donor region for transplantation. HDM is a typical object detection and classification problem that could benefit from deep learning. This study analyzed the accuracy of HDM by applying deep learning technology for object detection and reports the feasibility of automating HDM. The dataset for training and evaluation comprised 4492 enlarged hair scalp RGB images obtained from male hair-loss patients and the corresponding annotation data that contained the location information of the hair follicles present in the image and follicle-type information according to the number of hairs. EfficientDet, YOLOv4, and DetectoRS were used as object detection algorithms for performance comparison. The experimental results indicated that YOLOv4 had the best performance, with a mean average precision of 58.67.

GMRNet: A Novel Geometric Mean Relation Network for Few-Shot Very Similar Object Classification

With the widespread use of Deep Learning (DL), the use of DL has increased to provide a solution to the problem of object recognition and classification. In addition to classifying many different types of objects, the Deep Metrics Learning(DML) technique is effective in classifying objects that are visually very similar to each other. In this study, a novel Relation Network (RN) based DML has been designed to classify objects in two different datasets we created. We distinguished groups of objects that had a high degree of similarity to each other. These objects have been categorized using few-shot learning(FSL) since they are quite similar to one another. The impact of changing the number of classes and samples in the database on the network’s performance has been studied. It is shown how the network’s accuracy varies depending on the N-way (number of classes) and K-shots (number of samples) combinations used in its design. Additionally, the performance of the network has improved by an average of 15% thanks to the contribution of the recently introduced geometric mean module to the RN in our study. The accuracy rate of our recommended RN in screw and spare parts datasets is 96.1% and 92.3%, respectively. The first dataset consists of 1800 screw images with 18 classes, while the second dataset consists of 4100 spare parts images with 20 classes. The effectiveness of our method is expressed by the two datasets that we have extensively experimentally studied.

Deep Discriminative Feature Models (DDFMs) for Set Based Face Recognition and Distance Metric Learning.

This article introduces two methods that find compact deep feature models for approximating images in set based face recognition problems. The proposed method treats each image set as a nonlinear face manifold that is composed of linear components. To find linear components of the face manifold, we first split image sets into subsets containing face images which share similar appearances. Then, our first proposed method approximates each subset by using the center of the deep feature representations of images in those subsets. Centers modeling the subsets are learned by using distance metric learning. The second proposed method uses discriminative common vectors to represent image features in the subsets, and entire subset is approximated with an affine hull in this approach. Discriminative common vectors are subset centers that are projected onto a new feature space where the combined within-class variances coming from all subsets are removed. Our proposed methods can also be considered as distance metric learning methods using triplet loss function where the learned subcluster centers are the selected anchors. This procedure yields to applying distance metric learning to quantized data and brings many advantages over using classical distance metric learning methods. We tested proposed methods on various face recognition problems using image sets and some visual object classification problems. Experimental results show that the proposed methods achieve the state-of-the-art accuracies on the most of the tested image datasets.

A New Method Based on Deep Convolutional Neural Networks for Object Detection and Classification

Accurate object detection and classification has a broad application in industrial tasks, such as fabric defect and invoice detection. Previous state-of-the-art methods such as SSD and Faster-RCNN usually need to carefully adjust anchor box related hyper parameters and have poor performance in special fields with large object size/ratio variations and complex background texture. In this study, we proposed a new accurate, robust, and anchor-free method to handle automatic object detection and classification problems. First, we used the feature pyramid network (FPN), to merge the feature maps of different scales of features extracted from a convolutional neural network (CNN), which allowed easy and robust multi-scale feature fusion. Second, we built two subnets to generate candidate region proposals from the FPN outputs. followed by another CNN that determined the categories of the proposed regions from the two subnets.

A moving ISAR-object recognition using pi-sigma neural networks based on histogram of oriented gradient of edge

ABSTRACT Detection and classification with traditional neural networks methods such as multilayer perceptron (MLP), feed forward network and back propagation neural networks show several drawbacks including the rate of convergence and the incapacity facing the problems of size of the image especially for radar images. As a result, these methods are being replaced by other evolutional classification methods such as Higher Order Neural Networks (HONN) (Functional Link Artificial Neural Network (FLANN), Pi Sigma Neural Network (PSNN), Neural Network Product Unit (PUNN) and Neural Network of the Higher Order Processing Unit. So, in this paper, we address radar object detection and classification problems with a new strategy by using PSNN and a new proposed method HOGE for edges features extraction based on morphological operators and histogram of oriented gradient. Thus, in order to recognise radar object, we extract HOG features of the object region and classify our target with PSNN. The HOGE features vector is used as input of pi-sigma NN. The proposed method was tested and confirmed based on experiments through the use of 2D and 3D ISAR images.

A comparative analysis between late fusion of features approach and ensemble of multiple classifiers approach for image classification

SummaryIn recent times the late fusion of features approach for high‐level features extracted by multiple deep convolutional neural networks (DCNNs) has proven to be very effective in the computer vision field, especially for object classification problems. Pretrained DCNNs DenseNet‐121 and ResNet‐18 are retrained, keeping the number of output nodes equal to the number of classes present in the dataset. The last fully connected layers of these networks thereby get adapted to transform the high‐level features to a low‐dimensional feature map. Then these maps are fused to improve the performance of the model. On the other hand, an ensemble of multiple classifiers reduces the overfitting problem by combining multiple models prediction matrices. In this work, the prediction matrices of two Logsoftmax multiclass classifiers are combined. The feature maps for these two classifiers are extracted using pretrained DenseNet and ResNet. This study compares the late fusion of high‐level features approach and ensemble of multiple classifiers approach for object classification problems. Experimentation has been carried out on two benchmark datasets, such as CIFAR‐10 and CIFAR‐100, and it achieves 96.48% and 83.33% of test accuracy for ensemble of multiple classifiers and the late feature fusion approach. The proposed method has been compared with other deep architectures and datasets.

FEATURES OF METAHEURISTIC METHODS

The essence of tasks and classification, as well as the necessity of fulfillment of conditions and relations on the set are analyzed. The properties of the equivalence relation are specified. As elementary fragments are analyzed all edges of the graph. The conditions of attachment of an edge are the next: this edge is a ray of an existing star or has no common vertices with already constructed stars of coverage. Emphasis is placed on the lack of optimality of such a solution. The aim of this work is to construct metaheuristics for finding a suboptimal classification defined by a tolerance relation on a finite set. This approach allows one to construct partitions close to optimal sets in accordance with the relation of “proximity” of elements. Moreover, this relationship of proximity is not transitive. The proposed algorithms can find wide application in applied problems related to the problem of object classification by a number of attributes. Such problems often arise in the economic, social and technical sciences. The urgency of the fragmentary structure of the problem was emphasized. Specifies the ability to build classes by looking at the entire list of objects that are classified in a specific order. On the basis of fragmentary structure it is proposed to use evolutionary algorithm. The prospect of using a genetic algorithm to find the best classifications was evaluated. The step-by-step sequence of operations of the genetic algorithm with examples is shown: selection, crossing, mutation, selection. Examples of key operators are given, named crossovers and mutations. A detailed algorithm of the evolutionary model is clearly illustrated. The principle of action of the evolutionary fragmentary algorithm is described in detail. As a set of feasible solutions, a subset of maximal fragments on a given fragmentary structure is considered. The mechanism of checking the quality of the genetic algorithm on a fragmentary structure, which reduces to a lot of variants, is defined. The problem of finding optimal classifications on a finite set is investigated. It is shown that the problem of finding the optimal classification generated by the tolerance relation on a finite set is reduced to the problem of optimization on the set of permutations. A modification of the method of mixed jumping frogs for finding suboptimal solutions of the classification problem is proposed.

Landscape Similarity Analysis Using Texture Encoded Deep-Learning Features on Unclassified Remote Sensing Imagery

Convolutional neural networks (CNNs) are known for their ability to learn shape and texture descriptors useful for object detection, pattern recognition, and classification problems. Deeper layer filters of CNN generally learn global image information vital for whole-scene or object discrimination. In landscape pattern comparison, however, dense localized information encoded in shallow layers can contain discriminative information for characterizing changes across image local regions but are often lost in the deeper and non-spatial fully connected layers. Such localized features hold potential for identifying, as well as characterizing, process–pattern change across space and time. In this paper, we propose a simple yet effective texture-based CNN (Tex-CNN) via a feature concatenation framework which results in capturing and learning texture descriptors. The traditional CNN architecture was adopted as a baseline for assessing the performance of Tex-CNN. We utilized 75% and 25% of the image data for model training and validation, respectively. To test the models’ generalization, we used a separate set of imagery from the Aerial Imagery Dataset (AID) and Sentinel-2 for model development and independent validation. The classical CNN and the Tex-CNN classification accuracies in the AID were 91.67% and 96.33%, respectively. Tex-CNN accuracy was either on par with or outcompeted state-of-the-art methods. Independent validation on Sentinel-2 data had good performance for most scene types but had difficulty discriminating farm scenes, likely due to geometric generalization of discriminative features at the coarser scale. In both datasets, the Tex-CNN outperformed the classical CNN architecture. Using the Tex-CNN, gradient-based spatial attention maps (feature maps) which contain discriminative pattern information are extracted and subsequently employed for mapping landscape similarity. To enhance the discriminative capacity of the feature maps, we further perform spatial filtering, using PCA and select eigen maps with the top eigen values. We show that CNN feature maps provide descriptors capable of characterizing and quantifying landscape (dis)similarity. Using the feature maps histogram of oriented gradient vectors and computing their Earth Movers Distances, our method effectively identified similar landscape types with over 60% of target-reference scene comparisons showing smaller Earth Movers Distance (EMD) (e.g., 0.01), while different landscape types tended to show large EMD (e.g., 0.05) in the benchmark AID. We hope this proposal will inspire further research into the use of CNN layer feature maps in landscape similarity assessment, as well as in change detection.

Improving the Performance of Convolutional Neural Networks for Image Classification

As a high performance method for various image processing tasks, deep convolutional neural networks (CNNs) have reached impressive performances and absorbed considerable attention in the last few years. However, object classification on small size datasets for which a limited number of training images is available is still considered as an open problem. In this paper, we investigate a new method to effectively extract semantic image features. The proposed method which is based on CNNs boosts the performance of the object classification problem on small size dataset. To this end, a new method using image segmentation and CNNs is investigated. Our main goal is to increase the classification accuracy by first detecting and then extracting the main object of images. As training CNNs on small datasets does not yield to high performances because of millions of parameters to be learned, we propose using transfer learning strategy. Consequently, we first determine the main object of an image, and then we extract it. The extracted main object is used to tune the weights of the CNN in the training process. In this study, we employ a CNN that has been trained on the ImageNet dataset to reach mid-level image representation. Our experiments on Caltech-101 object dataset have shown that the proposed method substantially defeats other state-of-the-art methods.