Research on the Strawberry Recognition Algorithm Based on Deep Learning

Posture detection targeted towards providing assessments for the monitoring of health and welfare of pigs has been of great interest to researchers from different disciplines. Existing studies applying machine vision techniques are mostly based on methods using three-dimensional imaging systems, or two-dimensional systems with the limitation of monitoring under controlled conditions. Thus, the main goal of this study was to determine whether a two-dimensional imaging system, along with deep learning approaches, could be utilized to detect the standing and lying (belly and side) postures of pigs under commercial farm conditions. Three deep learning-based detector methods, including faster regions with convolutional neural network features (Faster R-CNN), single shot multibox detector (SSD) and region-based fully convolutional network (R-FCN), combined with Inception V2, Residual Network (ResNet) and Inception ResNet V2 feature extractions of RGB images were proposed. Data from different commercial farms were used for training and validation of the proposed models. The experimental results demonstrated that the R-FCN ResNet101 method was able to detect lying and standing postures with higher average precision (AP) of 0.93, 0.95 and 0.92 for standing, lying on side and lying on belly postures, respectively and mean average precision (mAP) of more than 0.93.

Disaster damage detection through synergistic use of deep learning and 3D point cloud features derived from very high resolution oblique aerial images, and multiple-kernel-learning

Deep learning target detection based on faster regions with convolutional neural network (Faster R-CNN) features has been applied in image processing successfully, however, it is rarely introduced to the field of hyperspectral image (HSI) target detection due to the tensor characteristics and the lack of training samples of HSI data. In this paper, the target detection based on Faster R-CNN is proposed to HSI with data set adjustment and parameter turning. As a typical tensor data, HSIs contain two-dimensional (2-D) spatial information and one dimensional (1-D) spectral information. It contains more information than ordinary images, and has unique advantages in the field of ground object and sea target detection. Therefore, the original HSI is firstly adjusted to the data set format required by the model, and the final Faster R-CNN sample data set can be achieved by combining the data set of Google Earth images. Next, a Faster R-CNN network suitable for HSI data could be built. Finally, to improve the accuracy of target detection, some parameters of Faster R-CNN would be tuned. The numerical results show that the method has the potential advantages of high precision and high speed in HSI target detection, and will have broad application prospects.

Co-trained convolutional neural networks for automated detection of prostate cancer in multi-parametric MRI.

The non-maximum suppression (NMS) algorithm, which merges neighboring bounding boxes as the detection result, is widely utilized in object detection methods. However, the traditional NMS algorithm is not suitable for text detection in natural scenes, especially for long texts and dense texts. In this paper, we observe that the coordinates of candidate bounding boxes are presented in skew distribution. On this observation, an improved NMS algorithm called SD-NMS (Skew Distribution NMS) is designed. First, the mode and the median of the coordinate set are counted to filter out the redundant bounding boxes. Then the left bounding boxes are merged for the location of the text regions. The SD-NMS method improves the detection ability of the model without extra model training and can be easily embedded in text detection methods. The experimental results show that our method obtains F-measure more than other NMS methods in public data sets ICDAR2015 and MSRATD500.

Deep learning has demonstrated significant capabilities for learning image features and presents many opportunities for agricultural automation. Deep neural networks typically require large and diverse training datasets to learn generalizable models. However, this requirement is challenging for applications in agricultural automation systems, since collecting and annotating large amount of training samples from filed crops and greenhouses is an expensive and complicated process due to the large diversity of crops, growth seasons and climate changes. This research proposed a new method for augmenting training dataset using synthesized images that preserves the background context and texture of the data object. A synthetic dataset of 1800 images was generated using a reference dataset and applying image processing techniques. As reference dataset 100 and for evaluating detection performance 230 real images of strawberry flowers were collected in greenhouses. Experimental results demonstrated that the suggested method provides improved performance when applied to the state-of-the-arts convolutional object detectors including Faster R-CNN, SSD, YOLOv3 and CenterNet for the task of strawberry flower detection in non-structured environment. The YOLOv3 w/darknet53 model achieved 46.84% boost in performance with average precision (AP) improved from 39.20% to 86.04% when applied augmentation using synthetic dataset. The AP of Faster R-CNN w/resnet50, SSD w/resnet50 and FPN and CenterNet w/hourglass52 models improved by 15.71, 18.42 and 22.24%, respectively. The Faster R-CNN w/resnet50 model provided most significant strawberry flower detection performance with AP 90.84%, which is higher than SSD w/resnet50 and FPN, YOLOv3 w/darknet53 and CenterNet w/hourglass52 models (88.56%, 86.04 % and 83.82%, respectively). Keywords: Flower detection, deep convolutional neural network, data augmentation, synthetic dataset.

The advent of deep learning techniques has ignited interest in medical image processing. The proposed work in this paper suggests one of the edge technologies in deep learning, which is recommended, based on a Radiomics feature extraction model for the effective detection of Kaposi sarcoma, a vascular skin lesion expression that indicates the most prevalent cancer in AIDS patients. This work investigates the role and impact of medical image fusion on deep feature learning based on ensemble learning in the medical domain. The model is crafted wherein the pre-built ResNet50 (Residual network) and Visual Geometry Group (VGG16) are fine-tuned and an ensemble learning approach is applied. The pre-defined CNN was incrementally regulated to determine the appropriate standards for classification efficiency improvements. Our findings show that layer-by-layer fine-tuning can improve the performance of middle and deep layers. This work would serve the purpose of masking and classification of skin lesion images, primarily sarcoma using an ensemble approach. Our proposed assisted framework could be deployed in assisting radiologists by classifying Kaposi sarcoma as well as other related skin lesion diseases, based on the positive classification findings.

A new approach for classification skin lesion based on transfer learning, deep learning, and IoT system

Detection of traffic signs plays an important role in autonomous driving, traffic surveillance and traffic safety. Previous research in Traffic Sign Detection (TSD) generally focused on traffic signs which are over the roads, the traffic signs on road surface have not been discussed. In this paper, we propose a road surface traffic sign detection system by applying convolutional neural network (CNN). The proposed system consists of two main stages: 1) a hybrid region proposal method to hypothesize the traffic sign locations by taking into account complementary information of color and edge; 2) feature extraction, classification, bounding box regression and non-maximum suppression by Fast R-CNN. Extensive experiments have been conducted using our field-captured dataset, demonstrating outstanding performance with regard to high recall and precision rate. The overall average precision (AP) is about 85.58%.

Non-maximum suppression (NMS) is an indispensable post-processing step in object detection. After a picture is detected by the target, a large number of redundant anchors will be obtained, so we need to go through NMS to process these repeated anchors. Its essence is to extract the target detection frame with high confidence and suppress the false detection frame with low confidence. The biggest problem in the NMS algorithm is that it completely removes adjacent low-confidence detection frames. In some images with dense targets, this is very likely to cause missed detection and false detection. Therefore, we proposed the Simple-NMS algorithm, which adds two new thresholds to the original NMS, and sets a constraint condition different from the original based on the new threshold, and does not change the complexity of the traditional NMS algorithm, which can be very good Improve the effectiveness of NMS missed targets. The new NMS algorithm is improved on the standard data sets PASAL VOC2007 and MS COCO. In addition, the algorithm is simple and efficient, and can be easily integrated into any other object detection process.

Some human diseases are recognized through of each type of White Blood Cell (WBC) count, so detecting and classifying each type is important for human healthcare. The main aim of this paper is to propose a computer-aided WBCs utility analysis tool designed, developed, and evaluated to classify WBCs into five types namely neutrophils, eosinophils, lymphocytes, monocytes, and basophils. Using a computer-artificial model reduces resource and time consumption. Various pre-trained deep learning models have been used to extract features, including AlexNet, Visual Geometry Group (VGG), Residual Network (ResNet), which belong to different taxonomy types of deep learning architectures. Also, Binary Border Collie Optimization (BBCO) is introduced as an updated version of Border Collie Optimization (BCO) for feature reduction based on maximizing classification accuracy. The proposed computer aid diagnosis tool merges transfer deep learning ResNet101, BBCO feature reduction, and Support Vector Machine (SVM) classifier to form a hybrid model ResNet101-BBCO-SVM an accurate and fast model for classifying WBCs. As a result, the ResNet101-BBCO-SVM scores the best accuracy at 99.21%, compared to recent studies in the benchmark. The model showed that the addition of the BBCO algorithm increased the detection accuracy, and at the same time, decreased the classification time consumption. The effectiveness of the ResNet101-BBCO-SVM model has been demonstrated and beaten in reasonable ratios in recent literary studies and end-to-end transfer learning of pre-trained models.

The histological assessment of glomeruli is fundamental for determining if a kidney is suitable for transplantation. The Karpinski score is essential to evaluate the need for a single or dual kidney transplant and includes the ratio between the number of sclerotic glomeruli and the overall number of glomeruli in a kidney section. The manual evaluation of kidney biopsies performed by pathologists is time-consuming and error-prone, so an automatic framework to delineate all the glomeruli present in a kidney section can be very useful. Our experiments have been conducted on a dataset provided by the Department of Emergency and Organ Transplantations (DETO) of Bari University Hospital. This dataset is composed of 26 kidney biopsies coming from 19 donors. The rise of Convolutional Neural Networks (CNNs) has led to a realm of methods which are widely applied in Medical Imaging. Deep learning techniques are also very promising for the segmentation of glomeruli, with a variety of existing approaches. Many methods only focus on semantic segmentation—which consists in segmentation of individual pixels—or ignore the problem of discriminating between non-sclerotic and sclerotic glomeruli, so these approaches are not optimal or inadequate for transplantation assessment. In this work, we employed an end-to-end fully automatic approach based on Mask R-CNN for instance segmentation and classification of glomeruli. We also compared the results obtained with a baseline based on Faster R-CNN, which only allows detection at bounding boxes level. With respect to the existing literature, we improved the Mask R-CNN approach in sliding window contexts, by employing a variant of the Non-Maximum Suppression (NMS) algorithm, which we called Non-Maximum-Area Suppression (NMAS). The obtained results are very promising, leading to improvements over existing literature. The baseline Faster R-CNN-based approach obtained an F-Measure of 0.904 and 0.667 for non-sclerotic and sclerotic glomeruli, respectively. The Mask R-CNN approach has a significant improvement over the baseline, obtaining an F-Measure of 0.925 and 0.777 for non-sclerotic and sclerotic glomeruli, respectively. The proposed method is very promising for the instance segmentation and classification of glomeruli, and allows to make a robust evaluation of global glomerulosclerosis. We also compared Karpinski score obtained with our algorithm to that obtained with pathologists’ annotations to show the soundness of the proposed workflow from a clinical point of view.

Deep learning has been applied to microseismic event detection over the past few years. However, it is still challenging to detect microseismic events from records with low signal-to-noise ratios (S/Ns). To achieve high accuracy of event detection in a low-S/N scenario, we have developed an end-to-end network that jointly performs denoising and classification tasks (JointNet) and applied it to fiber-optic distributed acoustic sensing (DAS) microseismic data. JointNet consists of 2D convolution layers that are suitable for extracting features (such as moveout and amplitude) of the dense DAS data. Moreover, JointNet uses a joint loss, rather than any intermediate loss, to simultaneously update the coupled denoising and classification modules. With the preceding advantages, JointNet is capable of simultaneously attenuating noise and preserving fine details of events and therefore improving the accuracy of event detection. We generate synthetic events and collect real background noise from a real hydraulic fracturing project and then expand them using data augmentation methods to yield sufficient training data sets. We train and validate the JointNet using training data sets of different S/Ns and compare it with the conventional classification networks visual geometry group (VGG) and deep VGG (DVGG). The results demonstrate the effectiveness of JointNet: it consistently outperforms the VGG and DVGG in all S/N scenarios and it has a superior capability to detect events, especially in a low-S/N scenario. Finally, we apply JointNet to detect microseismic events from the real DAS data acquired during hydraulic fracturing. JointNet successfully detects all manually detected events and has a better performance than VGG and DVGG.

Due to constant loads, gear wear, and harsh working conditions, gearboxes are subject to fault occurrences. Faults in the gearbox can cause damage to the engine components, create unnecessary noise, degrade efficiency, and impact power transfer. Hence, the detection of faults at an early stage is highly necessary. In this work, an effort was made to use transfer learning to identify gear failures under five gear conditions—healthy condition, 25% defect, 50% defect, 75% defect, and 100% defect—and three load conditions—no load, T1 = 9.6, and T2 = 13.3 Nm. Vibration signals were collected for various gear and load conditions using an accelerometer mounted on the casing of the gearbox. The load was applied using an eddy current dynamometer on the output shaft of the engine. The obtained vibration signals were processed and stored as vibration radar plots. Residual network (ResNet)‐50, GoogLenet, Visual Geometry Group 16 (VGG‐16), and AlexNet were the network models used for transfer learning in this study. Hyperparameters, including learning rate, optimizer, train‐test split ratio, batch size, and epochs, were varied in order to achieve the highest classification accuracy for each pretrained network. From the results obtained, VGG‐16 pretrained network outperformed all other networks with a classification accuracy of 100%.

Mango is a popular fruit for local consumption and export commodity. Currently, Indonesian mango export at 37.8 M accounted for 0.115% of world consumption. Pest and disease are the common enemies of mango that degrade the quality of mango yield. Specialized treatment in export destinations such as gamma-ray in Australia, or hot water treatment in Korea, demands pest-free and high-quality products. Artificial intelligence helps to improve mango pest and disease control. This paper compares the deep learning model on mango fruit pests and disease recognition. This research compares Visual Geometry Group 16 (VGG16), residual neural network 50 (ResNet50), InceptionResNet-V2, Inception-V3, and DenseNet architectures to identify pests and diseases on mango fruit. We implement transfer learning, adopt all pre-trained weight parameters from all those architectures, and replace the final layer to adjust the output. All the architectures are re-train and validated using our dataset. The tropical mango dataset is collected and labeled by a subject matter expert. The VGG16 model achieves the top validation and testing accuracy at 89% and 90%, respectively. VGG16 is the shallowest model, with 16 layers; therefore, the model was the smallest size. The testing time is superior to the rest of the experiment at 2 seconds for 130 testing images.