Mask Region-based Convolutional Neural Network Research Articles

Ship hull number detection and recognition under sparse samples

目的 舰船舷号检测识别是海面态势感知的关键技术，精准的舷号检测识别对海洋权益保护具有重要意义。但目前没有公开数据提供支持。为此，本文先构建了一个真实场景下的稀疏舰船舷号数据集（sparse ship hullnumber dataset in real scene，SSHN-RS），包含3 004幅舰船图像，共计11 328个舷号字符，覆盖了多国、各类、水平、倾斜、背景简单、背景复杂、光线不佳和被遮挡的舰船舷号样本，是一个具有挑战性的数据集。基于SSHN-RS，开展舰船舷号检测识别研究，其主要难点在于：1）样本稀疏，模型容易过拟合；2）舷号字符分布密集，网络难以充分提取各字符特征；3）部分字符存在嵌套区域和相似区域，网络会识别出大量冗余结果。针对上述难点，提出了一种基于多视角渐进式上下文解耦的舰船舷号检测识别算法。方法 首先，引入一个固定中心和最大化面积的随机透视变换技术，在不增加样本数量的前提下扩充舷号姿态，实现了数据增广，提升了模型的泛化能力；其次，提出了一个渐进式上下文解耦技术，先通过依次擦除舷号各字符生成一系列新样本，再利用特征提取网络提取和融合各样本的多尺度特征，不仅减少字符上下文信息对特征学习的干扰，而且再次增广了数据；最后，在测试阶段，提出了一个掩码间扰动抑制技术，先根据预测结果采用与渐进式上下文解耦技术类似的方法生成新样本并重新进行预测，再引入一个1维非极大值抑制技术去除预测结果中错误的冗余字符，输出最佳检测识别结果，进一步优化网络性能。结果 在SSHN-RS上采用主流实例分割算法进行定性和定量评估。在定量评估上，本文算法舷号的检测精确率、召回率、F值和识别率分别可达0.985 4，0.957 6，0.971 3，0.901 8，均优于其他算法。相比指标排名第2的算法，分别提高了4.51%，3.45%，3.97%，8.83%；在定性评估上，本文算法更适合舰船舷号检测识别任务，检测识别性能更高。此外，本文算法可以泛化到其他实例分割算法中，以经典算法Mask RCNN（mask region based convolutional neural network）为例，加入本文算法各模块后，各指标分别提升了9.82%，6.04%，7.80%，6.73%。结论 本文算法可以解决舷号检测识别任务中因样本稀疏、舷号分布密集、部分字符存在嵌套和相似性带来的问题，在主观和客观上均取得了最先进的性能，并且具有通用性。SSHN-RS可通过https://github.com/Bingchuan897/SSHN-RS获取。;Objective Ship hull number detection and recognition can be as the key technologies for marine awareness. It is essential for the preservation of maritime rights and interests. However，it is required to data-driven researches in support of ship hull number detection and recognition. Therefore，we develop a sparse ship hull number dataset in real scene （SSHN-RS），which contains 3 004 images with a total of 11 328 hull numbers. The challenging SSHN-RS dataset is featured of ship hull numbers of various countries，hull numbers of various types，horizontal hull numbers，inclined hull numbers，hull numbers with complex background，hull numbers with simple background，poorly illuminated hull numbers and partially occluded hull numbers. We carry out SSHN-RS-related research on ship hull number detection and recognition. The main challenges are required to be resolved on three aspects as following：1）the hull number samples are sparse， which causes over-fitting of the network，2）the features of hull number are densely distributed，which is challenged to learn some of the hull number characteristics fully，and 3）some hull number have its nested areas and a high degree of similarity，which is costly for large number of redundant results. Method To resolve these problems mentioned above，we demonstrate a ship hull number detection and recognition algorithm in terms of multi-view progressive context decoupling. First，a random perspective transformation technology with fixed center and maximized area is illustrated. To realize data augmentation and improve the generalization ability of the model，the hull number spatial attitude is extended without increasing the number of samples. Second，a progressive context decoupling technology is proposed. A series of new samples are first generated by sequentially erasing each character of the hull number，and the feature extraction network is then used to extract and fuse the multi-scale features of each sample. It can reduce the influence of feature-contextual information on feature learning and rich the data expansion. It can improve the feature expression ability effectively as well. Finally，in the testing stage，to generate new samples，and inputs the new samples into the testing network for prediction， an inter mask disturbance suppression technology is first focused on a method similar to the progressive context decoupling technology. At the same time，a one-dimensional non-maximum suppression technology is introduced that it can mainly process the hull number recognition results of each sample twice. To suppress the redundant mask in the detection and recognition results effectively，it can make each character of the hull number of all samples correspond to a recognition result. For the testing network，the output has only a set of optimal results. However，there is hull number noisy recognition in some samples. Therefore，the hull number recognition results of all samples are added on the original images，and the second one-dimensional non-maximum suppression technology is performed and processed on it. It can suppress the noises in some samples and outputs a set of optimal results. The post-processing module can optimize the detection and recognition performance further. Result The comparative experiment is mainly carried out on SSHN-RS. First，we conduct evaluation analysis on the general instance segmentation algorithms. Multi-view progressive context decoupling-based detection precision，recall，f-score and recognition rate of the ship number detection and recognition algorithm can reach 0. 985 4，0. 957 6， 0. 971 3 and 0. 901 8，which are improved by 4. 51%，3. 45%，3. 97% and 8. 83% respectively compared to the secondranked method. Second，the ablation experiments on the ship hull number detection and recognition algorithm are carried out in terms of multi-view progressive context decoupling. The experimental results show that the performance of ship hull number detection and recognition is improved whether each technology is used solely or mixed. Finally，to validate its effectiveness，we apply some popular modules of the ship hull number detection and recognition algorithm based on multiview progressive context decoupling to other general instance segmentation algorithms. Taking the classic algorithm mask region based convolutional neural network（Mask RCNN）as an example，the indexes have been improved by 9. 82%， 6. 04%，7. 80% and 6. 73% of each after our algorithm module is added. Conclusion Our SSHN-RS contains rich and effective ship hull number information，which can provide data support for ship hull number detection and recognition. The experimental results show that the multi-view progressive context decoupling-based ship hull number detection and recognition algorithm can optimize the detection and recognition performance of hull numbers due to sparse samples，dense character distribution，nested areas，and characters-between high similarity. This algorithm can be generalized to other deep learning based segmentation algorithms. The benchmarks-relevant are provided as a basis for future research on ship hull number detection and recognition. The dataset is available at https://github.com/Bingchuan897/SSHN-RS.

Extraction of Road Network in Urban Area from Orthophoto Using Deep Learning and Douglas-Peucker Post-Processing Algorithm

A Large-scale base map is needed by metropolitan cities such as Surabaya City for urban planning and smart city development. The most needed information from a large-scale base map is road geospatial information. Road network extraction is a challenging task for many reasons, including the heterogeneous attribute of the geometry and spectral, the complexity of objects that are difficult to model, and poor sensor data. Visual interpretation by operators is still a commonly used approach for extracting information from orthophoto. Interpretation accuracy depends on the skill and experience of the operator. So, there can be inconsistencies in the data generated by different operators. In recent years, the automatic extraction of road from orthophoto or VHR image has become an important and challenging research issue. Many recent studies have explored deep learning to improve the quality of building and road extraction. In this study, we applied a Mask Region-based Convolutional Neural Network (Mask R-CNN) model for the road network extraction using orthophoto in urban area in Surabaya City. The quality of the extracted geometry needs to be improved. Several post-processing strategies, including polygon regularization using Douglas-Peucker algorithm and polygon smoothing are designed to achieve optimal extraction results. The method produces a good performance for road extraction, the precision is 90.28%; recall 85.85%; F1-score 88.01%; and IoU 78.59%; and the overall accuracy is 95.25% and the kappa value is 90.5%.

Boundary Shape-Preserving Model for Building Mapping From High-Resolution Remote Sensing Images

The building is a critical urban element for developing smart cities. Building extraction research has experienced rapid development due to abundant remote sensing image data availability. Building extraction methods have recently focused on enhancing regional accuracy without obtaining refined building boundaries, and do not fit the concave-convex shape of the building and have significant sawtooth noise. This paper proposes a building shape-preserving framework to solve the imperfect boundary problem. To delineate the concave and convex building boundaries from remote sensing images, we combine coarse and fine-grained information with the instance segmentation method, Mask region-based convolutional neural network (R-CNN), which can further refine the indeterminate boundary points, and perform inward fitting of the building boundary. A regular boundary network is designed to learn the features of the edges and footprints to calculate the boundary loss that can smoothen the boundary noise, considering the imperfect boundaries and sawtooth noise in the extracted building boundary. Furthermore, to enhance the ability to evaluate the extracted building shape, a building footprint information evaluation algorithm, the footprint distance metric, is proposed to calculate the footprint distance between each building boundary. Comparing the building extraction results indicate that the proposed method achieves excellent quantitative evaluation scores. Experiments performed using Spacenet and WHU datasets to verify the effectiveness of the proposed method are detailed. The experimental results indicate that the algorithm utilizes the powerful edge extraction capability of the method to exhibit excellent performance in terms of extracting building edges from complex remote sensing scenes. Code is available at https://github.com/AnnaCUG/Boundary-shape-preserving-model.

Deep learning-assisted classification of calcaneofibular ligament injuries in the ankle joint

The classification of calcaneofibular ligament (CFL) injuries on magnetic resonance imaging (MRI) is time-consuming and subject to substantial interreader variability. This study explores the feasibility of classifying CFL injuries using deep learning methods by comparing them with the classifications of musculoskeletal (MSK) radiologists and further examines image cropping screening and calibration methods. The imaging data of 1,074 patients who underwent ankle arthroscopy and MRI examinations in our hospital were retrospectively analyzed. According to the arthroscopic findings, patients were divided into normal (class 0, n=475); degeneration, strain, and partial tear (class 1, n=217); and complete tear (class 2, n=382) groups. All patients were divided into training, validation, and test sets at a ratio of 8:1:1. After preprocessing, the images were cropped using Mask region-based convolutional neural network (R-CNN), followed by the application of an attention algorithm for image screening and calibration and the implementation of LeNet-5 for CFL injury classification. The diagnostic effects of the axial, coronal, and combined models were compared, and the best method was selected for outgroup validation. The diagnostic results of the models in the intragroup and outgroup test sets were compared with those results of 4 MSK radiologists of different seniorities. The mean average precision (mAP) of the Mask R-CNN using the attention algorithm for the left and right image cropping of axial and coronal sequences was 0.90-0.96. The accuracy of LeNet-5 for classifying classes 0-2 was 0.92, 0.93, and 0.92, respectively, for the axial sequences and 0.89, 0.92, and 0.90, respectively, for the coronal sequences. After sequence combination, the classification accuracy for classes 0-2 was 0.95, 0.97, and 0.96, respectively. The mean accuracies of the 4 MSK radiologists in classifying the intragroup test set as classes 0-2 were 0.94, 0.91, 0.86, and 0.85, all of which were significantly different from the model. The mean accuracies of the MSK radiologists in classifying the outgroup test set as classes 0-2 were 0.92, 0.91, 0.87, and 0.85, with the 2 senior MSK radiologists demonstrating similar diagnostic performance to the model and the junior MSK radiologists demonstrating worse accuracy. Deep learning can be used to classify CFL injuries at similar levels to those of MSK radiologists. Adding an attention algorithm after cropping is helpful for accurately cropping CFL images.

Detection and Reconstruction of Passion Fruit Branches via CNN and Bidirectional Sector Search.

Accurate detection and reconstruction of branches aid the accuracy of harvesting robots and extraction of plant phenotypic information. However, the complex orchard background and twisting growing branches of vine fruit trees make this challenging. To solve these problems, this study adopted a Mask Region-based convolutional neural network (Mask R-CNN) architecture incorporating deformable convolution to segment branches in complex backgrounds. Based on the growth posture, a branch reconstruction algorithm with bidirectional sector search was proposed to adaptively reconstruct the segmented branches obtained by an improved model. The average precision, average recall, and F1 scores of the improved Mask R-CNN model for passion fruit branch detection were found to be 64.30%, 76.51%, and 69.88%, respectively, and the average running time on the test dataset was 0.75 s per image, which is better than the compared model. We randomly selected 40 images from the test dataset to evaluate the branch reconstruction. The branch reconstruction accuracy, average error, average relative error of reconstructed diameter, and mean intersection-over-union (mIoU) were 88.83%, 1.98 px, 7.98, and 83.44%, respectively. The average reconstruction time for a single image was 0.38 s. This would promise the proposed method to detect and reconstruct plant branches under complex orchard backgrounds.

An experimental study on marine debris location and recognition using object detection

The large amount of debris in our oceans is a global problem that dramatically impacts marine fauna and flora. While a large number of human-based campaigns have been proposed to tackle this issue, these efforts have been deemed insufficient due to the insurmountable amount of existing litter. In response to that, there exists a high interest in the use of autonomous underwater vehicles (AUV) that may locate, identify, and collect this garbage automatically. To perform such a task, AUVs consider state-of-the-art object detection techniques based on deep neural networks due to their reported high performance. Nevertheless, these techniques generally require large amounts of data with fine-grained annotations. In this work, we explore the capabilities of the reference object detector Mask Region-based Convolutional Neural Networks for automatic marine debris location and classification in the context of limited data availability. Considering the recent CleanSea corpus, we pose several scenarios regarding the amount of available train data and study the possibility of mitigating the adverse effects of data scarcity with synthetic marine scenes. Our results achieve a new state of the art in the task, establishing a new reference for future research. In addition, it is shown that the task still has room for improvement and that the lack of data can be somehow alleviated, yet to a limited extent.

Automatic Disease Detection of Basal Stem Rot Using Deep Learning and Hyperspectral Imaging

Basal Stem Rot (BSR), a disease caused by Ganoderma boninense (G. boninense), has posed a significant concern for the oil palm industry, particularly in Southeast Asia, as it has the potential to cause substantial economic losses. The breeding programme is currently searching for G. boninense-resistant planting materials, which has necessitated intense manual screening in the nursery to track the progression of disease development in response to different treatments. The combination of hyperspectral image and machine learning approaches has a high detection potential for BSR. However, manual feature selection is still required to construct a detection model. Therefore, the objective of this study is to establish an automatic BSR detection at the seedling stage using a pre-trained deep learning model and hyperspectral images. The aerial view image of an oil palm seedling is divided into three regions in order to determine if there is any substantial spectral change across leaf positions. To investigate if the background images affect the performance of the detection, segmented images of the plant seedling have been automatically generated using a Mask Region-based Convolutional Neural Network (RCNN). Consequently, three models are utilised to detect BSR: a convolutional neural network that is 16 layers deep (VGG16) model trained on a segmented image; and VGG16 and Mask RCNN models both trained on the original images. The results indicate that the VGG16 model trained with the original images at 938 nm wavelength performed the best in terms of accuracy (91.93%), precision (94.32%), recall (89.26%), and F1 score (91.72%). This method revealed that users may detect BSR automatically without having to manually extract image attributes before detection.

Citrus Tree Crown Segmentation of Orchard Spraying Robot Based on RGB-D Image and Improved Mask R-CNN

Orchard spraying robots must visually obtain citrus tree crown growth information to meet the variable growth-stage-based spraying requirements. However, the complex environments and growth characteristics of fruit trees affect the accuracy of crown segmentation. Therefore, we propose a feature-map-based squeeze-and-excitation UNet++ (MSEU) region-based convolutional neural network (R-CNN) citrus tree crown segmentation method that intakes red–green–blue-depth (RGB-D) images that are pixel aligned and visual distance-adjusted to eliminate noise. Our MSEU R-CNN achieves accurate crown segmentation using squeeze-and-excitation (SE) and UNet++. To fully fuse the feature map information, the SE block correlates image features and recalibrates their channel weights, and the UNet++ semantic segmentation branch replaces the original mask structure to maximize the interconnectivity between feature layers, achieving a near-real time detection speed of 5 fps. Its bounding box (bbox) and segmentation (seg) AP50 scores are 96.6 and 96.2%, respectively, and the bbox average recall and F1-score are 73.0 and 69.4%, which are 3.4, 2.4, 4.9, and 3.5% higher than the original model, respectively. Compared with bbox instant segmentation (BoxInst) and conditional convolutional frameworks (CondInst), the MSEU R-CNN provides better seg accuracy and speed than the previous-best Mask R-CNN. These results provide the means to accurately employ autonomous spraying robots.

Effective Parameters to Design an Automatic Parking System

The automated parking system is an extensive branch of smart transport systems. The smartness of such systems is determined by different parameters such as parking maneuver planning. Coding this control system includes vehicle parking and understanding the environment. A high-quality classification mask has been used on each sample to analyze the automated vehicle parking parameters. Mask region-based convolutional neural networks (R-CNN) was taught using a small computational workload titled faster R-CNN that operates in five frames per second. In this paper, the rapidly-exploring random tree (RRT) method was used for routing the parking space and a nonlinear model predictive control (NMPC) controller was added to develop this system. We add the line detection algorithm commands to the mask R-CNN algorithm. The results can be useful to design a secure automatic parking system as well as a powerful perception system.

Precious Tree Pest Identification with Improved Instance Segmentation Model in Real Complex Natural Environments

It is crucial to accurately identify precious tree pests in a real, complex natural environment in order to monitor the growth of precious trees and provide growers with the information they need to make effective decisions. However, pest identification in real complex natural environments is confronted with several obstacles, including a lack of contrast between the pests and the background, the overlapping and occlusion of leaves, numerous variations in pest size and complexity, and a great deal of image noise. The purpose of the study was to construct a segmentation method for identifying precious tree pests in a complex natural environment. The backbone of an existing Mask region-based convolutional neural network was replaced with a Swin Transformer to improve its feature extraction capability. The experimental findings demonstrated that the suggested method successfully segmented pests in a variety of situations, including shaded, overlapped, and foliage- and branch-obscured pests. The proposed method outperformed the two competing methods, indicating that it is capable of accurately segmenting pests in a complex natural environment and provides a solution for achieving accurate segmentation of precious tree pests and long-term automatic growth monitoring.

Artificial intelligence based model for establishing the histopathological diagnostic of the cutaneous basal cell carcinoma

Abstract Introduction: Artificial intelligence (AI), a component of computer science, has the ability to process the multitude of medical data existing in the medical system around the world. The goal of our study is to build an AI model, based on Machine Learning, capable of assisting pathologists around the world in the diagnosis of the basal cell carcinoma of the skin. Material and Method: Our study is represented by the development of a Mask-RCNN (Mask Region-based Convolutional Neural Network) model, for the detection of cells with typical basal cell carcinoma tumoral changes. A number of 258 digitized histological images were used. The images emerged from Hematoxylin&Eosin stained pathology slides, diagnosed with cutaneous basal cell carcinoma between January 2018 and December 2021, at the Pathology Service of the Mureș County Clinical Hospital. Results: All the used images have the unique resolution of 2560x1920 pixels. For the learning process, we divided these images into two datasets: the learning dataset, representing 80% of the total images; and the test dataset, representing 20% of the total images. The AI model was trained using 1000 epochs with a learning rate of 0.00025 and only one classification category: basal cell carcinoma. Conclusions: The AI model successfully identified in 85% of the cases the areas with pathological changes present in the input images.

Application of Region-Based Convolutional Neural Network for Prompt Segmentation between Infected Cucumber Leaves and Healthy Cucumber Leaves

Plant diseased leaf image segmentation plays an important role in the plant disease detection through leaf symptoms, and early separation of infected and healthy plant leaves from each other can prevent horticulture loss. To achieve this goal, Region-Based Convolutional Neural Network (R-CNN) for prompt segmentation between infected cucumber leaves and healthy cucumber leaves was proposed and applied. A whole color cucumber leaf image is inputted into the convolutional neural network of the Mask R-CNN model, thereafter, the extracted features in their map are passed to the region proposal network for region proposals, the proposed regions of interest in their unaligned form are aligned before passing them to the fully connected layers in a fixed size feature map for the following actions: (1) bounding boxing, classification, and masking. The experimental results obtained in this work are on a par with the results obtained in literature, which demonstrates the effectiveness of the proposed method and high practical value for plant growth monitoring.

An efficient lung disease classification from X-ray images using hybrid Mask-RCNN and BiDLSTM

Lung diseases mainly affect the inner lining of the lungs causing complications in breathing, airway obstruction, and exhalation. Identifying lung diseases such as COVID-19, pneumonia, fibrosis, and tuberculosis at the earlier stage is a great challenge due to the availability of insufficient laboratory kits and image modalities. The rapid progression of the lung disease can be easily identified via Chest X-rays and this serves as a major boon for the terminally ill patients admitted to Intensive Care Units (ICU). To enhance the decision-making capability of the clinicians, a novel lung disease prediction framework is proposed using a hybrid bidirectional Long-Short-Term-Memory (BiDLSTM)-Mask Region-Based Convolutional Neural Network (Mask-RCNN) model. The Crystal algorithm is used to optimize the scalability and convergence issues in the Mask-RCNN model by hyperparameter tuning. The long-range dependencies for lung disease prediction are done using the BiDLSTM architecture which is connected to the fully connected layer of the Mask RCNN model. The efficiency of the proposed methodology is evaluated using three publicly accessible lung disease datasets namely the COVID-19 radiography dataset, Tuberculosis (TB) Chest X-ray Database, and National Institute of Health Chest X-ray Dataset which consists of the images of infected lung disease patients. The efficiency of the proposed technique is evaluated using different performance metrics such as Accuracy, Precision, Recall, F-measure, Specificity, confusion matrix, and sensitivity. The high accuracy obtained when comparing the proposed methodology with conventional techniques shows its efficiency of it in improving lung disease diagnosis.

Mask Region-Based Convolutional Neural Network (R-CNN) Based Image Segmentation of Rays in Softwoods

The current study aimed to verify the image segmentation ability of rays in tangential thin sections of conifers using artificial intelligence technology. The applied model was Mask region-based convolutional neural network (Mask R-CNN) and softwoods (viz. Picea jezoensis, Larix gmelinii, Abies nephrolepis, Abies koreana, Ginkgo biloba, Taxus cuspidata, Cryptomeria japonica, Cedrus deodara, Pinus koraiensis) were selected for the study. To take digital pictures, thin sections of thickness 10–15 μm were cut using a microtome, and then stained using a 1:1 mixture of 0.5% astra blue and 1% safranin. In the digital images, rays were selected as detection objects, and Computer Vision Annotation Tool was used to annotate the rays in the training images taken from the tangential sections of the woods. The performance of the Mask R-CNN applied to select rays was as high as 0.837 mean average precision and saving the time more than half of that required for Ground Truth. During the image analysis process, however, division of the rays into two or more rays occurred. This caused some errors in the measurement of the ray height. To improve the image processing algorithms, further work on combining the fragments of a ray into one ray segment, and increasing the precision of the boundary between rays and the neighboring tissues is required.

Automatic detection of early gastric cancer in endoscopy based on Mask region-based convolutional neural networks (Mask R-CNN)(with video)

The artificial intelligence (AI)-assisted endoscopic detection of early gastric cancer (EGC) has been preliminarily developed. The currently used algorithms still exhibit limitations of large calculation and low-precision expression. The present study aimed to develop an endoscopic automatic detection system in EGC based on a mask region-based convolutional neural network (Mask R-CNN) and to evaluate the performance in controlled trials. For this purpose, a total of 4,471 white light images (WLIs) and 2,662 narrow band images (NBIs) of EGC were obtained for training and testing. In total, 10 of the WLIs (videos) were obtained prospectively to examine the performance of the RCNN system. Furthermore, 400 WLIs were randomly selected for comparison between the Mask R-CNN system and doctors. The evaluation criteria included accuracy, sensitivity, specificity, positive predictive value and negative predictive value. The results revealed that there were no significant differences between the pathological diagnosis with the Mask R-CNN system in the WLI test (χ2 = 0.189, P=0.664; accuracy, 90.25%; sensitivity, 91.06%; specificity, 89.01%) and in the NBI test (χ2 = 0.063, P=0.802; accuracy, 95.12%; sensitivity, 97.59%). Among 10 WLI real-time videos, the speed of the test videos was up to 35 frames/sec, with an accuracy of 90.27%. In a controlled experiment of 400 WLIs, the sensitivity of the Mask R-CNN system was significantly higher than that of experts (χ2 = 7.059, P=0.000; 93.00% VS 80.20%), and the specificity was higher than that of the juniors (χ2 = 9.955, P=0.000, 82.67% VS 71.87%), and the overall accuracy rate was higher than that of the seniors (χ2 = 7.009, P=0.000, 85.25% VS 78.00%). On the whole, the present study demonstrates that the Mask R-CNN system exhibited an excellent performance status for the detection of EGC, particularly for the real-time analysis of WLIs. It may thus be effectively applied to clinical settings.

Deep learning-based postharvest strawberry bruise detection under UV and incandescent light

Bruising is one of the major defects of strawberries. Strawberry bruised areas are often visible as flattened, sunken, and discolored. Postharvest strawberries can be bruised by compression, impact, or vibration forces during harvesting, transportation, and packinghouse operations. These bruises make strawberries more vulnerable to rot or disease, thus shortening their shelf life. Bruises can also cause consumers to abandon purchase decisions. Therefore, inspecting strawberry bruises is an important procedure for commercial strawberry farms to guarantee the high quality of strawberries sent to the market. A novel technique was developed in this study to detect bruises on strawberry images captured by a color camera under incandescent and ultraviolet (UV) light. Mask Region-Based Convolutional Neural Networks (Mask R-CNN), a deep learning method, was utilized for the strawberry bruise detection. Four bruise severities were classified according to the estimated bruise ratio of every single strawberry (Minor Bruise, Light Bruise, Moderate Bruise, and Severe Bruise). The results showed that Mask R-CNN could accurately detect whether a strawberry is bruised or not under both lighting conditions. However, the strawberry bruise severity classification accuracy of the proposed machine vision technique was improved significantly under UV light compared to incandescent light. For whole strawberry bruise detection under UV light, the F1 score for detecting strawberry bruises was 0.99, and the highest F1 score for whole strawberry bruise severity classification was 0.92 for Minor Bruise. This strawberry bruise detection system can assist farmers in detecting bruised strawberries on the processing line or an automated strawberry harvester, improve the quality of strawberries that are sent to the market, and avoid shortening the shelf life of fresh strawberries.

Cosmic ray muon clustering for the MicroBooNE liquid argon time projection chamber using sMask-RCNN

In this article, we describe a modified implementation of Mask Region-based Convolutional Neural Networks (Mask-RCNN) for cosmic ray muon clustering in a liquid argon TPC and applied to MicroBooNE neutrino data. Our implementation of this network, called sMask-RCNN, uses sparse submanifold convolutions to increase processing speed on sparse datasets, and is compared to the original dense version in several metrics. The networks are trained to use wire readout images from the MicroBooNE liquid argon time projection chamber as input and produce individually labeled particle interactions within the image. These outputs are identified as either cosmic ray muon or electron neutrino interactions. We find that sMask-RCNN has an average pixel clustering efficiency of 85.9% compared to the dense network's average pixel clustering efficiency of 89.1%. We demonstrate the ability of sMask-RCNN used in conjunction with MicroBooNE's state-of-the-art Wire-Cell cosmic tagger to veto events containing only cosmic ray muons. The addition of sMask-RCNN to the Wire-Cell cosmic tagger removes 70% of the remaining cosmic ray muon background events at the same electron neutrino event signal efficiency. This event veto can provide 99.7% rejection of cosmic ray-only background events while maintaining an electron neutrino event-level signal efficiency of 80.1%. In addition to cosmic ray muon identification, sMask-RCNN could be used to extract features and identify different particle interaction types in other 3D-tracking detectors.

Comparison of Dairy Cow on Morphological Image Segmentation Model with Support Vector Machine Classification

Pattern recognition is viral in object recognition and classification, as it can cope with the complexity of problems related to the object of the image. For example, the category of dairy cows is essential for farmers to distinguish the quality of dairy cows for motherhood. The current problem with breeders is still using the selection process manually. If the selection process using the morphology of dairy cows requires the presence of computer vision. The purpose of this study is to make it easier for dairy farmers to choose the mothers to be farmed. This work uses several processes ranging from preprocessing, segmentation, and classification of images. This study used the classification of three segmentation algorithms, namely Canny, Mask Region-Based Convolutional Neural Networks (R-CNN), and K-Means. This method aims to compare the results of the segmentation algorithm model with SVM); the model is measured with accuracy, precision, recall, and F1 Score. The expected results get the most optimal model by using multiple resistant segmentation. The most optimal model testing achieved 90.29% accuracy, 92.49% precision, 89.39% recall, and 89.95% F1 Score with a training and testing ratio of 90:10. So the most optimal segmentation method uses the K-Means algorithm with a test ratio of 90:10.

Deep learning based instance segmentation of particle streaks and tufts

3D particle streak velocimetry (3D-PSV) and surface flow visualization using tufts both require the detection of curve segments, particle streaks or tufts, in images. We propose the use of deep learning based instance segmentation neural networks Mask region-based convolutional neural network (R-CNN) and Cascade Mask R-CNN, trained on fully synthetic data, to accurately identify, segment, and classify streaks and tufts. For 3D-PSV, we use the segmented masks and detected streak endpoints to volumetrically reconstruct flows even when the imaged streaks partly overlap or intersect. In addition, we use Mask R-CNN to segment images of tufts and classify the detected tufts according to their range of motion, thus automating the detection of regions of separated flow while at the same time providing accurate segmentation masks. Finally, we show a successful synthetic-to-real transfer by training only on synthetic data and successfully evaluating real data. The synthetic data generation is particularly suitable for the two presented applications, as the experimental images consist of simple geometric curves or a superposition of curves. Therefore, the proposed networks provide a general framework for instance detection, keypoint detection and classification that can be fine-tuned to the specific experimental application and imaging parameters using synthetic data.

Attention Mask R-CNN with edge refinement algorithm for identifying circulating genetically abnormal cells.

Recent studies have suggested that circulating tumor cells with abnormalities in gene copy numbers in mononuclear cell-enriched peripheral blood samples, such as circulating genetically abnormal cells (CACs), can be used as a non-invasive tool to detect patients with benign pulmonary nodules. These cells are identified through fluorescence signals counting by using 4-color fluorescence in situ hybridization (FISH) technology that exhibits high stability, sensitivity, and specificity. When FISH data are analyzed, the overlapping cells and fluorescence noise is a great challenge for identifying of CACs, thereby seriously affecting the efficiency of clinical diagnosis. To address this problem, in this study, we proposed an end-to-end FISH-based method (CACNET) for CAC identification. CACNET achieved nuclear segmentation and counted 4-color staining signals through improved Mask region-based convolutional neural network (R-CNN), followed by cell category (normal cell, deletion cell, gain cell, or CAC) according to pathological criteria. Firstly, the segmentation accuracy of overlapping nuclei was improved by adding an edge constraint head during training. Then, the interference of fluorescence noise was reduced by fusing non-local module to reconstruct the feature extraction network of Mask R-CNN. We trained and tested the proposed model on a dataset comprising 700 frames with 58,083 nuclei. The Accuracy, Sensitivity, and Specificity (overall performance metric for the algorithm) of CAC identification with CACNET were 94.06%, 92.1%, and 99.8%, respectively. Moreover, the developed method exhibited approximately identification speed of approximately 0.22 s per frames. The results showed that the proposed method outperformed the existing CAC identification methods, making it a promising approach for early screening of lung cancer.