International Journal of Pattern Recognition and Artificial Intelligence

The internet content increases exponentially day-by-day leading to the pop-up of irrelevant data while searching. Thus, the vast availability of web data requires curation to enhance the results of the search in relevance to searched topics. The proposed F-CapsNet deals with the content curation of web blog data through the novel integration of fuzzy logic with a machine learning algorithm. The input content to be curated is initially pre-processed and seven major features such as sentence position, bigrams, TF-IDF, cosine similarity, sentence length, proper noun score and numeric token are extracted. Then the fuzzy rules are applied to generate the extractive summary. After the extractive curation, the output is passed to the novel capsule network based deep auto-encoder where the abstractive summary is produced. The performance measures such as precision, recall, F1-score, accuracy and specificity are computed and the results are compared with the existing state-of-the-art methods. From the simulations performed, it has been proven that the proposed method for content curation is more efficient than any other method.

Aiming to solve the problem of tracking drift during movement, which was caused by the lack of discriminability of the feature information and the failure of a fixed template to adapt to the change of object appearance, the paper proposes an object tracking algorithm combining attention mechanism and correlation filter theory based on the framework of full convolutional Siamese neural networks. Firstly, the apparent information is processed by using the attention mechanism thought, where the object and search area features are optimized according to the spatial attention and channel attention module. At the same time, the cross-attention module is introduced to process the template branch and search area branch, respectively, which makes full use of the diversified context information of the search area. Then, the background perception correlation filter model with scale adaptation and learning rate adjustment is adopted into the model construction, using as a layer in the network model to realize the object template update. Finally, the optimal object location is determined according to the confidence map with similarity calculation. Experimental results show that the designed method in the paper can promote the object tracking performance under various challenging environments effectively; the success rate increases by 16.2%, and the accuracy rate increases by 16%.

Difficult object detection and class imbalance in object detection are the two main challenges faced by aerial image object detection. Difficult objects include small objects, objects of scale variation and objects with serious background interference. Class imbalances come from the number of different classes of objects and sampling of positive and negative samples. Due to these challenges, conventional object detection models usually cannot effectively detect objects in aerial images, especially in the balance between network speed and accuracy. In this paper, the YOLOv3 network structure was improved and an object detection method under the aerial visual scene (AVS-YOLO) was proposed. By introducing a type of densely connected feature pyramid strategy, a scale-aware attention module was constructed, considering both residual dense network blocks and the median-frequency-balancing mechanism. On this basis, an algorithm with ideal speed and accuracy for object detection is obtained. To verify the effectiveness of the algorithm, AVS-YOLO and YOLOv3 were both used to test the VisDrone-DET2019 and UAVDT. The experimental results show that the AP of AVS-YOLO increases by 6.22% and 5.09% on the VisDrone2019 and UAVDT datasets, respectively, compared with YOLOv3. In addition, the AP of AVS-YOLO is 1.82% higher than that of YOLOv4 on the VisDrone2019 dataset. In terms of detection speed, AVS-YOLO can process 31.8 frames per second on a single Nvidia GTX 2080Ti GPU, compared with 44.1 frames per second for YOLOv3. Compared with the other one-stage network in the field of object detection, AVS-YOLO currently achieves the state-of-the-art performance with similar calculation amount on this dataset.

In recent years, machine learning models have achieved magnificent success in many industrial applications, but most of them are black boxes. It is crucial to understand why such predictions are made in many critical areas such as medicine, financial markets, and auto driving. In this paper, we propose Coco, a novel interpretation method which can interpret any binary classifier by assigning each feature an importance value for a particular prediction. We first adopt MixUp method to generate reasonable perturbations, then apply these perturbations with constraints to obtain counterfactual instances and finally compute a comprehensive metric on these instances to estimate the importance of each feature. To demonstrate the effectiveness of Coco, we conduct extensive experiments on several datasets. The results show our method achieves better performance in identifying the most important features compared with the state-of-the-art interpretation methods, including Shap and Lime.

Modern model-based reinforcement learning methods for high-dimensional inputs often incorporate an unsupervised learning step for dimensionality reduction. The training objective of these unsupervised learning methods often leverages only static inputs such as reconstructing observations. These representations are combined with predictor functions for simulating rollouts to navigate the environment. We advance this idea by taking advantage of the fact that we navigate dynamic environments with visual stimulus and create a representation that is specifically designed with control and actions in mind. We propose to learn a feature map that is maximally predictable for a predictor function. This results in representations that are well suited for the task of planning, where the predictor is used as a forward model. To this end, we introduce a new way of learning this representation along with the prediction function, a system we dub Latent Representation Prediction Network (LARP). The prediction function is used as a forward model for a search on a graph in a viewpoint-matching task, and the representation learned to maximize predictability is found to outperform other representations. The sample efficiency and overall performance of our approach are shown to rival standard reinforcement learning methods, and our learned representation transfers successfully to unseen environments.

Adversarial learning stability has an important influence on the generated image quality and convergence process in generative adversarial networks (GANs). Training dataset (real data) noise and the balance of game players have an impact on adversarial learning stability. In the gradient backpropagation of the discriminator, the noise samples increase the gradient variance. It can increase the uncertainty in the network convergence progress and affect stability. In the two-player zero-sum game, the game ability of the generator and discriminator is unbalanced. Generally, the game ability of the generator is weaker than that of the discriminator, which affects the stability. To improve the stability, an antinoise learning and coalitional game generative adversarial network (ANL-CG GAN) is proposed, which achieves this goal through the following two strategies. (i) In the real data loss function of the discriminator, an effective antinoise learning method is designed, which can improve the gradient variance and network convergence uncertainty. (ii) In the zero-sum game, a generator coalitional game module is designed to enhance its game ability, which can improve the balance between the generator and discriminator via a coalitional game strategy. To verify the performance of this model, the generated results of the designed GAN and other GAN models in CELEBA and CIFAR10 are compared and analyzed. Experimental results show that the novel GAN can improve adversarial learning stability, generate image quality, and reduce the number of training epochs.

In image retrieval, highlighting target object and reducing the influence of background noise remains challenging. To address this problem, we propose a novel weighting method that aggregates deep convolutional features based on filtering, called filtering on spatial channel weighting (FSCW) factors, to represent image contents, and utilize it for image retrieval. There are three main contributions of this study. First, the designed filter can effectively remove the influence of background noise. Second, we propose a new channel selection and spatial weighting method, which can accurately distinguish target object from the background noise. Finally, we designed a new channel weighting strategy to suppress intra-image visual burstiness. Experimental results on benchmark datasets demonstrate that the proposed method effectively enhances discriminative power and outperforms some existing state-of-the-art methods in terms of the mAP metric. Furthermore, the proposed method is superior to some existing algorithms in distinguishing background noise and target object.

The predictions of characters/text/digits from the handwritten images have made the research community spotlight towards recognition. There are enormous applications and ambiguity that made prediction possible with Deep Learning (DL) approaches. Primarily, there are four necessary steps to be carried out with handwriting prediction. First, consideration of a dataset that is more appropriate for DL validation an inefficient manner. Here, Special Database 1 and Special Database 2 are used, which are combined and modified by the National Institute of Standards and Technology (NIST). Next is pre-processing of input handwritten digit recognition data by data normalization, extraction of efficient features which provides better prediction accuracy. The proposed idea uses pixel values as features with the analysis of hyper-parameters to enhance near-human performance. With SVM, non-linear and linear models are built to extract the appropriate features for further processing. The features are separate and placed over the Bag of Features (BoF), which is used by the next processing stage. Finally, a novel Convolutional Neural Network (CNN) is by built modifying the network structure with Orthogonal Learning Particle Swarm Optimization (CNN-OLPSO). This modification is adopted for evolutionarily optimizing the number of hyper-parameters. This proposed optimizer predicts the optimal values from the fitness computation and shows better efficiency when compared to various other conventional approaches. The novelty which relies on CNN adoption is to endeavor a suitable path towards digitalization and preserve the handwritten structure and help automatic feature extraction using CNN by offering better computation accuracy. The optimization approach helps to avoid over-fitting and under-fitting issues. Here, metrics like accuracy, elapsed time, recall, precision, and $F$-measure are evaluated. The results of CNN-OLPSO give better accuracy, reduced error rate and better execution time (s) compared to other existing methods. Thus, the proposed model shows better tradeoff in the recognition rate of handwritten digits.

Leaf recognition has been an important research field of image recognition in the recent past. However, traditional leaf recognition methods can be easily affected by environments and cannot realize multi-leaf recognition under a complex background in real time. In this work, we present a real-time leaf recognition method based on image segmentation and feature recognition. First, we denoise the input of a leaf image, performing a leaf segmentation with an improved FCN network model, and then optimize the contour edge with a CRF algorithm to get a leaf segmentation image. Second, we extract the content features of the segmented leaf image with an Inception-V2 network model to get a feature map of the leaf image. Third, we input the feature map into an RPN network to obtain a set of regional candidate frames and then integrate the feature map and the information of candidate frames in a RoI Pooling layer, which can extract the feature map of a candidate frame area and scale it to a fixed-size feature map. Finally, we send the feature map to a fully connected layer to classify each preselection box content through the calculation of preselection feature maps, and then obtain the final accurate position of the prediction box by utilizing a bounding box regression. The experimental results show that the proposed method can achieve multi-leaf recognitions with high accuracy and fast speed under complex environments in real time.

Coastline detection using a Gaussian Mixture Model (GMM) applied to synthetic aperture radar (SAR) imagery is usually inaccurate due to the inherent noise of SAR data. In addition, the traditional active counter model is sensitive to the initial position of the contour line and requires a large number of iterations to converge to a solution. In this study, we first used the GMM algorithm to segment the SAR images and obtain a coarse land and sea segmentation map. This map is then used as the initial position for a subsequent active contour model. The K distribution was introduced into the local statistical active contour model to better model the SAR image. The Gaussian distribution-based local active contour model and the algorithm detailed in this paper were used to perform coastline extraction experiments on four SAR images. Four GF-3 SAR images with different modes were collected to validate the efficiency of the proposed method. The experimental results show that the coastline extraction methods from SAR images based on the GMM algorithm and the K distribution-based local statistical active contour model (LKDACM) overcame the shortcomings of the traditional active contour model to accurately and quickly detect coastlines, thus enabling the detection of coastline changes.

Extraction of agricultural parcels from high-resolution satellite imagery is an important task in precision agriculture. Here, we present a semi-automatic approach for agricultural parcel detection that achieves high accuracy and efficiency. Unlike the techniques presented in previous literatures, this method is pixel based, and it exploits the properties of a spectral angle mapper (SAM) to develop customized operators to accurately derive the parcels. The main steps of the method are sample selection, textural analysis, spectral homogenization, SAM, thresholding, and region growth. We have systematically evaluated the algorithm proposed on a variety of images from Gaofen-1 wide field of view (GF-1 WFV), Resource 1-02C (ZY1-02C), and Gaofen-2 (GF-2) to aerial image; the accuracies are 99.09% of GF-1 WFV, 84.42% of ZY1-02C, 96.51% and 92.18% of GF-2, and close to 100% of aerial image; these results demonstrated its accuracy and robustness.

In the process of multimodal image fusion, how to improve the visual effect after the image fused, while taking into account the protection of energy and the extraction of details, has attracted more and more attention in recent years. Based on the research of visual saliency and the final action-level measurement of the base layer, a multimodal image fusion method based on a guided filter is proposed in this paper. Firstly, multi-scale decomposition of a guided filter is used to decompose the two source images into a small-scale layer, large-scale layer and base layer. The fusion rule of the maximum absolute value is adopted in the small-scale layer, the weight fusion rule based on regular visual parameters is adopted in the large-scale layer and the fusion rule based on activity-level measurement is adopted in the base layer. Finally, the fused three scales are laminated into the final fused image. The experimental results show that the proposed method can improve the image edge processing and visual effect in multimodal image fusion.

Detection process of airborne targets may be thought simple because of the incompatible nature of aircraft, choppers, UAVs, and drones regarding clear sky background. When changes in the background are considered, brightness variation of the sky complicates the process. Changes in the shapes and types of clouds add another challenge to the process. Tracking process directly depends on the detection process and type of the data stream. The practical systems used for video detection and tracking of airborne targets are manual, and manual structures have some drawbacks compared to automatic structures. For video surveillance, guidance, regional security, and defense applications in dense environments, automatic detection and tracking process may be an obligation rather than preference. In this study, an automatic detection and tracking algorithm for video streams of airborne targets is proposed. A land-based moving camera captures the video data, and not only the flying objects but probably also the camera are in motion. Although the detection and tracking of moving objects via moving sensors is a relatively arduous task, this is the prevalent case in real-life scenarios. Video detection and tracking systems have one or more moving video sensors, while one or more flying air vehicles are in operation area. The proposed algorithm includes an image processing stage for detection and a tracking stage for initiation and continuation. An assessment study has been conducted for the actual video data and found that the proposed method yields successful results for detection, track formation, and continuation processes.

Detection of violent human behavior is necessary for public safety and monitoring. However, it demands constant human observation and attention in human-based surveillance systems, which is a challenging task. Autonomous detection of violent human behavior is therefore essential for continuous uninterrupted video surveillance. In this paper, we propose a novel method for violence detection and localization in videos using the fusion of spatio-temporal features and attention model. The model consists of Fusion Convolutional Neural Network (Fusion-CNN), spatio-temporal attention modules and Bi-directional Convolutional LSTMs (BiConvLSTM). The Fusion-CNN learns both spatial and temporal features by combining multi-level inter-layer features from both RGB and Optical flow input frames. The spatial attention module is used to generate an importance mask to focus on the most important areas of the image frame. The temporal attention part, which is based on BiConvLSTM, identifies the most significant video frames which are related to violent activity. The proposed model can also localize and discriminate prominent regions in both spatial and temporal domains, given the weakly supervised training with only video-level classification labels. Experimental results evaluated on different publicly available benchmarking datasets show the superior performance of the proposed model in comparison with the existing methods. Our model achieves the improved accuracies (ACC) of 89.1%, 99.1% and 98.15% for RWF-2000, HockeyFight and Crowd-Violence datasets, respectively. For CCTV-FIGHTS dataset, we choose the mean average precision (mAp) performance metric and our model obtained 80.7% mAp.

In order to overcome the phenomenon of aliasing mode in empirical mode decomposition (EMD), an improved EMD method based on frequency rotation factor and vector operation is proposed. First, fast Fourier transform (FFT) is used to determine the frequency components contained in the original signal. Second, Hilbert transform is applied to transform the real plane analysis into complex plane analysis, then the amplitude frequency product of the signal is satisfied with adaptive full decomposed by frequency inversion and vector operation. Third, the traditional EMD algorithm is used to extract the lowest frequency mode in the complex frequency domain while it is the highest frequency in the time domain (because the frequency inversion operation is performed in the previous step), and finally the original mode of the real plane is restored by forward frequency rotation and inverse vector operation, so as to complete all modal decomposition through this loop algorithm. The simulation results show that the method can effectively separate the aliasing problem and extract the useful signals with low time cost.

PIWI-interacting RNA (PiRNA) was discovered in 2006 and is expected to become a new biomarker for diagnosis and prognosis of various diseases. The purpose of this study is to explore functions of piRNAs and identify cancer subtypes on the basis of the pattern of transcriptome and somatic mutation data. A total of 285 510 SNPs in piRNAs and genes, which might affect piRNA biogenesis or piRNA targets binding were identified. Significant co-expression networks of piRNAs were then constructed separately for 12 major types of cancer. Finally, mutational matrices were mapped to piRNA network, propagated, and clustered for identification of cancer-related piRNAs and cancer subtypes. Findings showed that subtypes of three types of cancer (COAD, STAD and UCEC), which are significantly associated with survival were identified. Analysis of differentially expressed piRNAs in UCEC subtypes showed that piRNA function is closely related to cancer hallmarks “Enabling Replicative Immortality” and contributes to initiation of cancer.