Application of local fully Convolutional Neural Network combined with YOLO v5 algorithm in small target detection of remote sensing image.

This exploration primarily aims to jointly apply the local FCN (fully convolution neural network) and YOLO-v5 (You Only Look Once-v5) to the detection of small targets in remote sensing images. Firstly, the application effects of R-CNN (Region-Convolutional Neural Network), FRCN (Fast Region-Convolutional Neural Network), and R-FCN (Region-Based-Fully Convolutional Network) in image feature extraction are analyzed after introducing the relevant region proposal network. Secondly, YOLO-v5 algorithm is established on the basis of YOLO algorithm. Besides, the multi-scale anchor mechanism of Faster R-CNN is utilized to improve the detection ability of YOLO-v5 algorithm for small targets in the image in the process of image detection, and realize the high adaptability of YOLO-v5 algorithm to different sizes of images. Finally, the proposed detection method YOLO-v5 algorithm + R-FCN is compared with other algorithms in NWPU VHR-10 data set and Vaihingen data set. The experimental results show that the YOLO-v5 + R-FCN detection method has the optimal detection ability among many algorithms, especially for small targets in remote sensing images such as tennis courts, vehicles, and storage tanks. Moreover, the YOLO-v5 + R-FCN detection method can achieve high recall rates for different types of small targets. Furthermore, due to the deeper network architecture, the YOL v5 + R-FCN detection method has a stronger ability to extract the characteristics of image targets in the detection of remote sensing images. Meanwhile, it can achieve more accurate feature recognition and detection performance for the densely arranged target images in remote sensing images. This research can provide reference for the application of remote sensing technology in China, and promote the application of satellites for target detection tasks in related fields.

Automated segmentation by deep learning of loose connective tissue fibers to define safe dissection planes in robot-assisted gastrectomy.

The prediction of anatomical structures within the surgical field by artificial intelligence (AI) is expected to support surgeons' experience and cognitive skills. We aimed to develop a deep-learning model to automatically segment loose connective tissue fibers (LCTFs) that define a safe dissection plane. The annotation was performed on video frames capturing a robot-assisted gastrectomy performed by trained surgeons. A deep-learning model based on U-net was developed to output segmentation results. Twenty randomly sampled frames were provided to evaluate model performance by comparing Recall and F1/Dice scores with a ground truth and with a two-item questionnaire on sensitivity and misrecognition that was completed by 20 surgeons. The model produced high Recall scores (mean 0.606, maximum 0.861). Mean F1/Dice scores reached 0.549 (range 0.335-0.691), showing acceptable spatial overlap of the objects. Surgeon evaluators gave a mean sensitivity score of 3.52 (with 88.0% assigning the highest score of 4; range 2.45-3.95). The mean misrecognition score was a low 0.14 (range 0-0.7), indicating very few acknowledged over-detection failures. Thus, AI can be trained to predict fine, difficult-to-discern anatomical structures at a level convincing to expert surgeons. This technology may help reduce adverse events by determining safe dissection planes.

Hybrid Gradient Descent Grey Wolf Optimizer for Optimal Feature Selection.

Feature selection is the process of decreasing the number of features in a dataset by removing redundant, irrelevant, and randomly class-corrected data features. By applying feature selection on large and highly dimensional datasets, the redundant features are removed, reducing the complexity of the data and reducing training time. The objective of this paper was to design an optimizer that combines the well-known metaheuristic population-based optimizer, the grey wolf algorithm, and the gradient descent algorithm and test it for applications in feature selection problems. The proposed algorithm was first compared against the original grey wolf algorithm in 23 continuous test functions. The proposed optimizer was altered for feature selection, and 3 binary implementations were developed with final implementation compared against the two implementations of the binary grey wolf optimizer and binary grey wolf particle swarm optimizer on 6 medical datasets from the UCI machine learning repository, on metrics such as accuracy, size of feature subsets, F-measure, accuracy, precision, and sensitivity. The proposed optimizer outperformed the three other optimizers in 3 of the 6 datasets in average metrics. The proposed optimizer showed promise in its capability to balance the two objectives in feature selection and could be further enhanced.

Block-cyclic stochastic coordinate descent for deep neural networks.

We present a stochastic first-order optimization algorithm, named block-cyclic stochastic coordinate descent (BCSC), that adds a cyclic constraint to stochastic block-coordinate descent in the selection of both data and parameters. It uses different subsets of the data to update different subsets of the parameters, thus limiting the detrimental effect of outliers in the training set. Empirical tests in image classification benchmark datasets show that BCSC outperforms state-of-the-art optimization methods in generalization leading to higher accuracy within the same number of update iterations. The improvements are consistent across different architectures and datasets, and can be combined with other training techniques and regularizations.

Uncorrelated feature encoding for faster image style transfer.

Recent image style transfer methods use a pre-trained convolutional neural network as their feature encoder. However, the pre-trained network is not optimal for image style transfer but rather for image classification. Furthermore, they require time-consuming feature alignment to consider the existing correlation among channels of the encoded feature map. In this paper, we propose an end-to-end learning method that optimizes both encoder and decoder networks for style transfer task and relieves the computational complexity of the existing correlation-aware feature alignment. First, we performed end-to-end learning that updates not only decoder but also encoder parameters for the task of image style transfer in the network training phase. Second, in addition to the previous style and content losses, we use uncorrelation loss, i.e., the total correlation coefficient among responses of encoder channels. Our uncorrelation loss allows the encoder network to generate a feature map of channels without correlation. Subsequently, our method results in faster forward processing with only a light-weighted transformer of correlation-unaware feature alignment. Moreover, our method drastically reduced the channel redundancy of the encoded feature during the network training process. This provides us a possibility to perform channel elimination with negligible degradation in generated style quality. Our method is applicable to multiple scaled style transfer by using the cascade network scheme and allows a user to control style strength through the usage of a content-style trade-off parameter.

SPLASH: Learnable activation functions for improving accuracy and adversarial robustness.

We introduce SPLASH units, a class of learnable activation functions shown to simultaneously improve the accuracy of deep neural networks while also improving their robustness to adversarial attacks. SPLASH units have both a simple parameterization and maintain the ability to approximate a wide range of non-linear functions. SPLASH units are: (1) continuous; (2) grounded (f(0)=0); (3) use symmetric hinges; and (4) their hinges are placed at fixed locations which are derived from the data (i.e. no learning required). Compared to nine other learned and fixed activation functions, including ReLU and its variants, SPLASH units show superior performance across three datasets (MNIST, CIFAR-10, and CIFAR-100) and four architectures (LeNet5, All-CNN, ResNet-20, and Network-in-Network). Furthermore, we show that SPLASH units significantly increase the robustness of deep neural networks to adversarial attacks. Our experiments on both black-box and white-box adversarial attacks show that commonly-used architectures, namely LeNet5, All-CNN, Network-in-Network, and ResNet-20, can be up to 31% more robust to adversarial attacks by simply using SPLASH units instead of ReLUs. Finally, we show the benefits of using SPLASH activation functions in bigger architectures designed for non-trivial datasets such as ImageNet.

A New Image Classification Approach via Improved MobileNet Models with Local Receptive Field Expansion in Shallow Layers.

Because deep neural networks (DNNs) are both memory-intensive and computation-intensive, they are difficult to apply to embedded systems with limited hardware resources. Therefore, DNN models need to be compressed and accelerated. By applying depthwise separable convolutions, MobileNet can decrease the number of parameters and computational complexity with less loss of classification precision. Based on MobileNet, 3 improved MobileNet models with local receptive field expansion in shallow layers, also called Dilated-MobileNet (Dilated Convolution MobileNet) models, are proposed, in which dilated convolutions are introduced into a specific convolutional layer of the MobileNet model. Without increasing the number of parameters, dilated convolutions are used to increase the receptive field of the convolution filters to obtain better classification accuracy. The experiments were performed on the Caltech-101, Caltech-256, and Tubingen animals with attribute datasets, respectively. The results show that Dilated-MobileNets can obtain up to 2% higher classification accuracy than MobileNet.

Improved object recognition using neural networks trained to mimic the brain's statistical properties.

The current state-of-the-art object recognition algorithms, deep convolutional neural networks (DCNNs), are inspired by the architecture of the mammalian visual system, and are capable of human-level performance on many tasks. As they are trained for object recognition tasks, it has been shown that DCNNs develop hidden representations that resemble those observed in the mammalian visual system (Razavi and Kriegeskorte, 2014; Yamins and Dicarlo, 2016; Gu and van Gerven, 2015; Mcclure and Kriegeskorte, 2016). Moreover, DCNNs trained on object recognition tasks are currently among the best models we have of the mammalian visual system. This led us to hypothesize that teaching DCNNs to achieve even more brain-like representations could improve their performance. To test this, we trained DCNNs on a composite task, wherein networks were trained to: (a) classify images of objects; while (b) having intermediate representations that resemble those observed in neural recordings from monkey visual cortex. Compared with DCNNs trained purely for object categorization, DCNNs trained on the composite task had better object recognition performance and are more robust to label corruption. Interestingly, we found that neural data was not required for this process, but randomized data with the same statistical properties as neural data also boosted performance. While the performance gains we observed when training on the composite task vs the "pure" object recognition task were modest, they were remarkably robust. Notably, we observed these performance gains across all network variations we studied, including: smaller (CORNet-Z) vs larger (VGG-16) architectures; variations in optimizers (Adam vs gradient descent); variations in activation function (ReLU vs ELU); and variations in network initialization. Our results demonstrate the potential utility of a new approach to training object recognition networks, using strategies in which the brain - or at least the statistical properties of its activation patterns - serves as a teacher signal for training DCNNs.

Decoding spectro-temporal representation for motor imagery recognition using ECoG-based brain-computer interfaces.

One of the challenges in brain-computer interface systems is obtaining motor imagery recognition from brain activities. Brain-signal decoding robustness and system performance improvement during the motor imagery process are two of the essential issues in brain-computer interface research. In conventional approaches, ineffective decoding of features and high complexity of algorithms often lead to unsatisfactory performance. A novel method for the recognition of motor imagery tasks is developed based on employing a modified S-transforms for spectro-temporal representation to characterize the behavior of electrocorticogram activities. A classifier is trained by using a support vector machine, and an optimized wrapper approach is applied to guide selection to implement the representation selection obtained. A channel selection algorithm optimizes the wrapper approach by adding a cross-validation step, which effectively improves the classification performance. The modified S-transform can accurately capture event-related desynchronization/event-related synchronization phenomena and can effectively locate sensorimotor rhythm information. The optimized wrapper approach used in this scheme can effectively reduce the feature dimension and improve algorithm efficiency. The method is evaluated on a public electrocorticogram dataset with a recognition accuracy of 98% and an information transfer rate of 0.8586 bit/trial. To verify the effect of the channel selection, both electrocorticogram and electroencephalogram data are experimentally analyzed. Furthermore, the computational efficiency of this scheme demonstrates its potential for online brain-computer interface systems in future cognitive tasks.

3D-Convolutional Neural Network with Generative Adversarial Network and Autoencoder for Robust Anomaly Detection in Video Surveillance.

As the surveillance devices proliferate, various machine learning approaches for video anomaly detection have been attempted. We propose a hybrid deep learning model composed of a video feature extractor trained by generative adversarial network with deficient anomaly data and an anomaly detector boosted by transferring the extractor. Experiments with UCSD pedestrian dataset show that it achieves 94.4% recall and 86.4% precision, which is the competitive performance in video anomaly detection.

A novel method for detecting morphologically similar crops and weeds based on the combination of contour masks and filtered Local Binary Pattern operators.

BACKGROUND: Weeds are a major cause of low agricultural productivity. Some weeds have morphological features similar to crops, making them difficult to discriminate. RESULTS: We propose a novel method using a combination of filtered features extracted by combined Local Binary Pattern operators and features extracted by plant-leaf contour masks to improve the discrimination rate between broadleaf plants. Opening and closing morphological operators were applied to filter noise in plant images. The images at 4 stages of growth were collected using a testbed system. Mask-based local binary pattern features were combined with filtered features and a coefficient k. The classification of crops and weeds was achieved using support vector machine with radial basis function kernel. By investigating optimal parameters, this method reached a classification accuracy of 98.63% with 4 classes in the "bccr-segset" dataset published online in comparison with an accuracy of 91.85% attained by a previously reported method. CONCLUSIONS: The proposed method enhances the identification of crops and weeds with similar appearance and demonstrates its capabilities in real-time weed detection.

Diagnostic potential of automated tractography in progressive supranuclear palsy variants.

BACKGROUND: Microstructural white matter integrity captured by diffusion-tensor imaging (DTI) is significantly more affected in progressive supranuclear palsy-Richardson's syndrome (PSP-RS) compared to PSP-parkinsonism (PSP-P). OBJECTIVES: To characterize the microstructural integrity of large fascicular bundles using standardized probabilistic tractography and combine it with previously established DTI- and volumetric measures of subcortical brain structures in order to evaluate its diagnostic properties for the differentiation of PSP- RS, PSP-P and Parkinson's disease (PD). METHODS: DTI metrics as well as volumes of subcortical brain regions, acquired by 3T MRI of patients with PSP-RS (n = 15), PSP-P (n = 13), and a mean disease duration of 2.7 ± 1.8 years, were quantified by probabilistic tractography as well as a validated infratentorial atlas and compared to PD (n = 18) and healthy controls (n = 20). Classification accuracy of MRI measures was tested by consecutive linear discriminant analyses. RESULTS: DTI metrics of the anterior thalamic radiation, the corticospinal tract, the superior longitudinal fasciculus, the bundles of the corpus callosum and cingulate, the dentatorubrothalamic tract as well as volumes of the dorsal midbrain, globus pallidus and thalamus were significantly altered in PSP-RS and to a lesser extent in PSP-P compared to PD and healthy controls. Linear discriminant analysis identified DTI metrics of the dentatorubrothalamic tract and the anterior thalamic radiation as well as the volume of the dorsal midbrain to classify correctly 91.3% of PSP-RS, PSP-P and PD patients. CONCLUSIONS: Observer-independent investigations of microstructural integrity of major fiber bundles improved existing MRI processing strategies to differentiate PSP-P from PSP-RS and PD, in their early disease stages.

Weighted discriminative collaborative competitive representation for robust image classification.

Collaborative representation-based classification (CRC) is a famous representation-based classification method in pattern recognition. Recently, many variants of CRC have been designed for many classification tasks with the good classification performance. However, most of them ignore the inter-class pattern discrimination among the class-specific representations, which is very critical for strengthening the pattern discrimination of collaborative representation (CR). In this article, we propose a novel CR approach for image classification, called weighted discriminative collaborative competitive representation (WDCCR). The proposed WDCCR designs the discriminative and competitive collaborative representation among all the classes by fully considering the class information. On the one hand, we incorporate two discriminative constraints into the unified WDCCR model. Both constraints are the competitive class-specific representation residuals and the pairs of class-specific representations for each query sample. On the other hand, the constraint of the weighted categorical representation coefficients is introduced into the proposed model for further enhancing the power of discriminative and competitive representation. In the weighted constraint, we assume that the different classes of each query sample should have less contribution to the representation with the small representation coefficients, and then two types of weight factors are designed to constrain the representation coefficients. Furthermore, the robust WDCCR (R-WDCCR) is proposed with l1-norm representation fidelity for recognizing noisy images. Extensive experiments on six image data sets demonstrate the effective and robust superiorities of the proposed WDCCR and R-WDCCR over the related state-of-the-art representation-based classification methods.

RFDCR: Automated brain lesion segmentation using cascaded random forests with dense conditional random fields.

Segmentation of brain lesions from magnetic resonance images (MRI) is an important step for disease diagnosis, surgical planning, radiotherapy and chemotherapy. However, due to noise, motion, and partial volume effects, automated segmentation of lesions from MRI is still a challenging task. In this paper, we propose a two-stage supervised learning framework for automatic brain lesion segmentation. Specifically, in the first stage, intensity-based statistical features, template-based asymmetric features, and GMM-based tissue probability maps are used to train the initial random forest classifier. Next, the dense conditional random field optimizes the probability maps from the initial random forest classifier and derives the whole tumor regions referred as the region of interest (ROI). In the second stage, the optimized probability maps are further intergraded with features from the intensity-based statistical features and template-based asymmetric features to train subsequent random forest, focusing on classifying voxels within the ROI. The output probability maps will be also optimized by the dense conditional random fields, and further used to iteratively train a cascade of random forests. Through hierarchical learning of the cascaded random forests and dense conditional random fields, the multimodal local and global appearance information is integrated with the contextual information, and the output probability maps are improved layer by layer to finally obtain optimal segmentation results. We evaluated the proposed method on the publicly available brain tumor datasets BRATS 2015 & BRATS 2018, as well as the ischemic stroke dataset ISLES 2015. The results have shown that our framework achieves competitive performance compared to the state-of-the-art brain lesion segmentation methods. In addition, contralateral difference and skewness were identified as the important features in the brain tumor and ischemic stroke segmentation tasks, which conforms to the knowledge and experience of medical experts, further reflecting the reliability and interpretability of our framework.

TbCAPs: A toolbox for co-activation pattern analysis.

Functional magnetic resonance imaging provides rich spatio-temporal data of human brain activity during task and rest. Many recent efforts have focussed on characterising dynamics of brain activity. One notable instance is co-activation pattern (CAP) analysis, a frame-wise analytical approach that disentangles the different functional brain networks interacting with a user-defined seed region. While promising applications in various clinical settings have been demonstrated, there is not yet any centralised, publicly accessible resource to facilitate the deployment of the technique. Here, we release a working version of TbCAPs, a new toolbox for CAP analysis, which includes all steps of the analytical pipeline, introduces new methodological developments that build on already existing concepts, and enables a facilitated inspection of CAPs and resulting metrics of brain dynamics. The toolbox is available on a public academic repository at https://c4science.ch/source/CAP_Toolbox.git. In addition, to illustrate the feasibility and usefulness of our pipeline, we describe an application to the study of human cognition. CAPs are constructed from resting-state fMRI using as seed the right dorsolateral prefrontal cortex, and, in a separate sample, we successfully predict a behavioural measure of continuous attentional performance from the metrics of CAP dynamics (R â€‹= â€‹0.59).

A reference library for assigning protein subcellular localizations by image-based machine learning.

Confocal micrographs of EGFP fusion proteins localized at key cell organelles in murine and human cells were acquired for use as subcellular localization landmarks. For each of the respective 789,011 and 523,319 optically validated cell images, morphology and statistical features were measured. Machine learning algorithms using these features permit automated assignment of the localization of other proteins and dyes in both cell types with very high accuracy. Automated assignment of subcellular localizations for model tail-anchored proteins with randomly mutated C-terminal targeting sequences allowed the discovery of motifs responsible for targeting to mitochondria, endoplasmic reticulum, and the late secretory pathway. Analysis of directed mutants enabled refinement of these motifs and characterization of protein distributions in within cellular subcompartments.

Test-retest reliability of FreeSurfer automated hippocampal subfield segmentation within and across scanners.

The human hippocampus is vulnerable to a range of degenerative conditions and as such, accurate in vivo measurement of the hippocampus and hippocampal substructures via neuroimaging is of great interest for understanding mechanisms of disease as well as for use as a biomarker in clinical trials of novel therapeutics. Although total hippocampal volume can be measured relatively reliably, it is critical to understand how this reliability is affected by acquisition on different scanners, as multiple scanning platforms would likely be utilized in large-scale clinical trials. This is particularly true for hippocampal subregional measurements, which have only relatively recently been measurable through common image processing platforms such as FreeSurfer. Accurate segmentation of these subregions is challenging due to their small size, magnetic resonance imaging (MRI) signal loss in medial temporal regions of the brain, and lack of contrast for delineation from standard neuroimaging procedures. Here, we assess the test-retest reliability of the FreeSurfer automated hippocampal subfield segmentation procedure using two Siemens model scanners (a Siemens Trio and Prismafit Trio upgrade). T1-weighted images were acquired for 11 generally healthy younger participants (two scans on the Trio and one scan on the Prismafit). Each scan was processed through the standard cross-sectional stream and the recently released longitudinal pipeline in FreeSurfer v6.0 for hippocampal segmentation. Test-retest reliability of the volumetric measures was examined for individual subfields as well as percent volume difference and Dice overlap among scans and intra-class correlation coefficients (ICC). Reliability was high in the molecular layer, dentate gyrus, and whole hippocampus with the inclusion of three time points with mean volume differences among scans less than 3%, overlap greater than 80%, and ICC >0.95. The parasubiculum and hippocampal fissure showed the least improvement in reliability with mean volume difference greater than 5%, overlap less than 70%, and ICC scores ranging from 0.78 to 0.89. Other subregions, including the CA regions, were stable in their mean volume difference and overlap (<5% difference and >75% respectively) and showed improvement in reliability with the inclusion of three scans (ICC â€‹> â€‹0.9). Reliability was generally higher within scanner (Trio-Trio), however, Trio-Prismafit reliability was also high and did not exhibit an obvious bias. These results suggest that the FreeSurfer automated segmentation procedure is a reliable method to measure total as well as hippocampal subregional volumes and may be useful in clinical applications including as an endpoint for future clinical trials of conditions affecting the hippocampus.

Test-retest of automated segmentation with different motion correction strategies: A comparison of prospective versus retrospective methods.

Test-retest of automated image segmentation algorithms (FSL FAST, FSL FIRST, and FREESURFER) are computed on magnetic resonance images from 12 unsedated children aged 9.4±2.6 years ([min,max] â€‹= â€‹[6.5 years, 13.8 years]) using different approaches to motion correction (prospective versus retrospective). The prospective technique, PROMO MPRAGE, dynamically estimates motion using specially acquired navigator images and adjusts the remaining acquisition accordingly, whereas the retrospective technique, MPnRAGE, uses a self-navigation property to retrospectively estimate and account for motion during image reconstruction. To increase the likelihood and range of motions, participants heads were not stabilized with padding during repeated scans. When motion was negligible both techniques had similar performance. When motion was not negligible, the automated image segmentation and anatomical labeling software tools showed the most consistent performance with the retrospectively corrected MPnRAGE technique (≥80% volume overlaps for 15 of 16 regions for FIRST and FREESURFER, with greater than 90% volume overlaps for 12 regions with FIRST and 11 regions with FREESURFER). Prospectively corrected MPRAGE with linear view-ordering also demonstrated lower performance than MPnRAGE without retrospective motion correction.

Using Machine Learning and Structural Neuroimaging to Detect First Episode Psychosis: Reconsidering the Evidence.

Despite the high level of interest in the use of machine learning (ML) and neuroimaging to detect psychosis at the individual level, the reliability of the findings is unclear due to potential methodological issues that may have inflated the existing literature. This study aimed to elucidate the extent to which the application of ML to neuroanatomical data allows detection of first episode psychosis (FEP), while putting in place methodological precautions to avoid overoptimistic results. We tested both traditional ML and an emerging approach known as deep learning (DL) using 3 feature sets of interest: (1) surface-based regional volumes and cortical thickness, (2) voxel-based gray matter volume (GMV) and (3) voxel-based cortical thickness (VBCT). To assess the reliability of the findings, we repeated all analyses in 5 independent datasets, totaling 956 participants (514 FEP and 444 within-site matched controls). The performance was assessed via nested cross-validation (CV) and cross-site CV. Accuracies ranged from 50% to 70% for surfaced-based features; from 50% to 63% for GMV; and from 51% to 68% for VBCT. The best accuracies (70%) were achieved when DL was applied to surface-based features; however, these models generalized poorly to other sites. Findings from this study suggest that, when methodological precautions are adopted to avoid overoptimistic results, detection of individuals in the early stages of psychosis is more challenging than originally thought. In light of this, we argue that the current evidence for the diagnostic value of ML and structural neuroimaging should be reconsidered toward a more cautious interpretation.