Region Proposal Network Research Articles

A Multi-Scale Target Detection Method Using an Improved Faster Region Convolutional Neural Network Based on Enhanced Backbone and Optimized Mechanisms.

Currently, existing deep learning methods exhibit many limitations in multi-target detection, such as low accuracy and high rates of false detection and missed detections. This paper proposes an improved Faster R-CNN algorithm, aiming to enhance the algorithm's capability in detecting multi-scale targets. This algorithm has three improvements based on Faster R-CNN. Firstly, the new algorithm uses the ResNet101 network for feature extraction of the detection image, which achieves stronger feature extraction capabilities. Secondly, the new algorithm integrates Online Hard Example Mining (OHEM), Soft non-maximum suppression (Soft-NMS), and Distance Intersection Over Union (DIOU) modules, which improves the positive and negative sample imbalance and the problem of small targets being easily missed during model training. Finally, the Region Proposal Network (RPN) is simplified to achieve a faster detection speed and a lower miss rate. The multi-scale training (MST) strategy is also used to train the improved Faster R-CNN to achieve a balance between detection accuracy and efficiency. Compared to the other detection models, the improved Faster R-CNN demonstrates significant advantages in terms of mAP@0.5, F1-score, and Log average miss rate (LAMR). The model proposed in this paper provides valuable insights and inspiration for many fields, such as smart agriculture, medical diagnosis, and face recognition.

Enhanced recognition of insulator defects on power transmission lines via proposal-based detection model with integrated improvement methods

Deep learning-driven transmission line inspection is a critical area for smart power grid development. Despite advances in deep learning for insulator defect detection, challenges remain in model robustness and adaptability for the varying real-world adaptability, especially for insignificant defects in complex backgrounds. This study presents a comprehensive improvement strategy for detecting insulators and cross-scale broken defects on transmission lines, employing a proposal-based detection model. The model introduces a holistic pipeline of improved methods, including backbone modification, anchor box scale recalibration, and improvements in Region of Interest (RoI) downsampling alignment and Intersection over Union (IoU) loss function. Various backbone networks, including convolutional network (ConvNet) and Vision Transformer (ViT) structures, are constructed and integrated with attention modules, specifically designed to amplify the perception of insulators and defective regions. The geometric scale of anchor boxes is reconstructed using a developed clustering algorithm, considering the elongated characteristics of insulator strings to improve the adaptability of anchor boxes. Bilinear interpolation is utilized to mitigate spatial misalignment issues during the downsampling process of Region Proposal Network (RPN)-based proposals. The experimental results indicate that the improved models with the Swin Transformer (Swin-T) backbone framework achieve the mean Average Precision (mAP)@0.5 of 88.42% and mAP@0.7 of 60.52%, with a defect recall rate of 81.94%. Additionally, the improved IoU loss function contributes to the performance of the model at higher IoU thresholds. The results of this study contribute to the further development of defect detection frameworks for power vision applications.

Improved faster region convolutional neural network algorithm for UAV target detection in complex environment

With the development of artificial intelligence technology and Unmanned Aerial Vehicle (UAV) technology, the traditional UAV target detection methods are difficult to achieve high detection accuracy in complex and changeable scenes. The paper selects Faster Region Convolutional Neural Network (Faster R-CNN) as the basic algorithm due to its high scalability and excellent classification performance. The convolution mode of the convolution layer in the network structure is adjusted to deformable convolution. Additionally, a hybrid attention mechanism module is added to combine channel attention and spatial attention modules behind the output layer of the Faster R-CNN network. Finally, the Soft Non-Maximum Suppression (Soft-NMS) method is selected to combine the improved Faster R-CNN algorithm of multiple individuals into the final target detection model. The performance of the improved Faster R-CNN algorithm and the original Faster R-CNN algorithm was verified through the VisDrone 2018-DET dataset and the full class average accuracy Mean Average Precision (mAP), accuracy and precision. The accuracy and logarithm of loss values of the improved Faster R-CNN algorithm and the original Faster R-CNN algorithm's Region Proposal Network (RPN) were 0.985, 0.981, 0.018, and 0.052, respectively. The target detection model based on the hybrid attention mechanism of ResNet-50 network fusion Spartial Attention Module (SAM) and Channel Attention Module (CAM) had the best classification performance, with the accuracy, precision and overall average accuracy of 25.8 %, 24.6 % and 22.7 %, respectively. The results are helpful to improve the target detection ability of UAV in complex environment, and contribute to the development of target detection technology in the future.

Mutual learning with memory for semi-supervised pest detection.

Effectively monitoring pest-infested areas by computer vision is essential in precision agriculture in order to minimize yield losses and create early scientific preventative solutions. However, the scale variation, complex background, and dense distribution of pests bring challenges to accurate detection when utilizing vision technology. Simultaneously, supervised learning-based object detection heavily depends on abundant labeled data, which poses practical difficulties. To overcome these obstacles, in this paper, we put forward innovative semi-supervised pest detection, PestTeacher. The framework effectively mitigates the issues of confirmation bias and instability among detection results across different iterations. To address the issue of leakage caused by the weak features of pests, we propose the Spatial-aware Multi-Resolution Feature Extraction (SMFE) module. Furthermore, we introduce a Region Proposal Network (RPN) module with a cascading architecture. This module is specifically designed to generate higher-quality anchors, which are crucial for accurate object detection. We evaluated the performance of our method on two datasets: the corn borer dataset and the Pest24 dataset. The corn borer dataset encompasses data from various corn growth cycles, while the Pest24 dataset is a large-scale, multi-pest image dataset consisting of 24 classes and 25k images. Experimental results demonstrate that the enhanced model achieves approximately 80% effectiveness with only 20% of the training set supervised in both the corn borer dataset and Pest24 dataset. Compared to the baseline model SoftTeacher, our model improves mAP @0.5 (mean Average Precision) at 7.3 compared to that of SoftTeacher at 4.6. This method offers theoretical research and technical references for automated pest identification and management.

Unified multimodal fusion transformer for few shot object detection for remote sensing images

Object detection is a fundamental computer vision task with wide applications in remote sensing, but traditional methods strongly rely on large annotated datasets which are difficult to obtain, especially for novel object classes. Few-shot object detection (FSOD) aims to address this by using detectors to learn from very limited labeled data. Recent work fuse multi-modalities like image–text pairs to tackle data scarcity but require external region proposal network (RPN) to align cross-modal pairs which leads to a bias towards base classes and insufficient cross-modal contextual learning. To address these problems, we propose a unified multi-modal fusion transformer (UMFT), which extracts visual features from ViT and textual encodings from BERT to align multi-modal representations in an end-to-end manner. Specifically, affinity-guided fusion (AFM) captures semantically related image–text pairs by modeling their affinity relationships to selectively combine most informative pairs. In addition, cross-modal correlation module (CCM) captures discriminative inter-modal patterns between image and text representations and filters out unrelated features to enhance cross-modal alignment. By leveraging AFM to focus on semantic relationships and CCM to refine inter-modal features, the model better aligns multimodal data without RPN. These representations are passed to detection decoder that predicts bounding boxes, probabilities of class and cross-modal attributes. Evaluation of UMFT on benchmark datasets NWPU VHR-10 and DIOR demonstrated its ability to leverage limited image–text training data via dynamic fusion, achieving new state-of-the-art mean average precision (mAP) for few-shot object detection. Our code will be publicly available at https://github.com/abdullah-azeem/umft.

A Novel Liver Tumor segmentation of Adverse Propagation Advanced Swin Transformer Network with Mask region-based convolutional neural networks

The diagnosis and treatment of liver diseases from computed tomography (CT) images is an indispensable task for segmentation of Liver & its tumours. Due to the uneven presence, fuzzy borders, diverse densities, shapes and sizes of lesions segmentation of liver & its tumour is a difficult task. At this point we mainly focused on deep learning algorithms for segmenting liver and its tumour from abdominal CT scan images thereafter minimising the time & energy used for a liver diseases diagnosis This study aims to classify and segment liver tumors using a novel deep learning-based model. A Mask region-based convolutional neural network (Mask R-CNN) model is proposed for multiorgan segmentation to aid esophageal radiation treatment. Due to the fact that organ boundaries may be fuzzy and organ shapes are various, original Mask R-CNN works well on natural image segmentation while leaves something to be desired on the multiorgan segmentation task. Addressing it, the advantages of this method are threefold: (1) a ROI (region of interest) generation method is presented in the RPN (region proposal network) which is able to utilize multiscale semantic features. (2) A prebackground classification subnetwork is integrated to the original mask generation branch to improve the precision of multiorgan segmentation. The segmented image is added to the Adverse Propagation Advanced Swin Transformer Network (APESTNet) to prevent overfitting. The proposed model is based on CT volume slices of patients with liver tumors and evaluated on the public 3D dataset IRCADB01.The proposed model dataset is split into training as 60% and testing as 40 % for classification. The proposed model's accuracy and recall are 99.23 % and 98.24 %, respectively.

Face recognition with occluded face using improve intersection over union of region proposal network on Mask region convolutional neural network

Face recognition entails detecting and identifying facial attributes. Mask region convolutional neural network (R-CNN) method is a prominent approach, while prior research predominantly delved into refining loss functions and perfecting object and face detection, recognizing, and identifying faces using imperfect data remained relatively unexplored. This study focuses on an occluded dataset comprising Indonesian faces, wherein 'occluded' denotes facial data that lacks complete visibility-encompassing instances where objects obscure faces or are partially cropped. This investigation involves a deliberate experiment that tailors the intersection over union (IoU) of the region proposal network (RPN) to suit the nuances of occluded Indonesian faces, thereby augmenting accuracy in recognition and segmentation tasks. The innovation IoU in the strategic utilization of Anchors, which involves the exclusion of anchors falling beyond the image borders to optimize computational efficiency. The outcomes of this research are striking; it showcases a remarkable 14.75%, 10.9%, and 12.97% surge based on mean average precision (mAP), mean average recall (mAR), and F1-Scores compared to the conventional Mask R-CNN approach. Notably, our proposed model elevates the average accuracy by 10% to 15% and decreases running time by 21%, a noteworthy enhancement compared to the preceding model. This progress is substantiated by validation utilizing 300 instances dataset, reinforcing the robustness of our approach.

An Adaptive Region Proposal Network With Progressive Attention Propagation for Tiny Person Detection From UAV Images

An Adaptive Region Proposal Network With Progressive Attention Propagation for Tiny Person Detection From UAV Images

Defect detection of printed circuit board based on adaptive key-points localization network

Many deep neural networks (DNNs) have been applied in the defect detection of products. Due to the irregular and small defects on printed circuit boards (PCB), it is difficult for the DNN-based defect detection models to achieve good detection performance. In this paper, a new DNN, adaptive key point localization network (AKPLNet) is proposed for PCB defect detection. Firstly, residual pyramid heat mapping network (RFHNet) that is composed of ResNet50_FPN and thermodynamic mechanism (TM), is used to perform multi-scale feature extraction and defect location. Secondly, an adaptive tree structure region proposal network (AT-RPN) based on tree structure Parzen estimation is proposed to obtain the predicted regions of the target, which reduces the need for large number of priori knowledge during the detection process. Finally, a key point regression algorithm is proposed to locate defects accurately. The defect detection performance of AKPLNet is validated on two PCB datasets. The mean average precision (mAP) of AKPLNet reaches 96.9% and 99.0% on PCB-Master dataset with the color images and DeepPCB-Master dataset with the grayscale images, improving 2.1% and 2.3% compared with Yolov7, respectively. The testing results demonstrate that AKPLNet achieves the better detection accuracy than those state-of-the-art methods.

TSAF-Net: a rotated two-stage Cnaphalocrocis medinalis damage detection method based on anchor-free arbitrary-oriented proposal network.

Cnaphalocrocis medinalis (C.medinalis) is an agricultural pest with recurrent outbreaks. The investigation into automated pest and disease detection technology holds significant value for in-field surveys. Current generic detection methods are inadequate due to arbitrary orientations and a wide range of aspect ratios in damage symptoms. To tackle these issues, we put forward a rotated two-stage detection method for in-field C.medinalis surveys. This method relies on an anchor-free rotated region proposal network (AF-R2PN), bypassing the need for hyper-parameter optimization induced by predefined anchor boxes. An in-field C.medinalis dataset is constructed during on-site pest surveys to validate the effectiveness of our method. The experimental results show that our method can accomplish 80% average precision (AP), surpassing the corresponding horizontal detector by 2.3%. The visualization results of our work showcase its exceptional localization capability over generic detection methods, facilitating inspection by plant protectors. Meanwhile, our proposed method outperforms other state-of-the-art rotated detection algorithms. The AF-R2PN module can generate superior arbitrary-oriented proposals even with a decreased number of proposals, balancing inference speed and detection performance among other rotated two-stage methods. The proposed method exhibits superiority in detecting C. medinalis damage under complex field conditions. It provides greater practical applicability during in-field surveys, enhancing their efficiency and coverage. The findings hold significance for pest and disease monitoring, providing important technical support for agricultural production. © 2024 Society of Chemical Industry.

Study of a Deep Convolution Network with Enhanced Region Proposal Network in the Detection of Cancerous Lung Tumors.

A deep convolution network that expands on the architecture of the faster R-CNN network is proposed. The expansion includes adapting unsupervised classification with multiple backbone networks to improve the Region Proposal Network in order to improve accuracy and sensitivity in detecting minute changes in images. The efficiency of the proposed architecture is investigated by applying it to the detection of cancerous lung tumors in CT (computed tomography) images. This investigation used a total of 888 images from the LUNA16 dataset, which contains CT images of both cancerous and non-cancerous tumors of various sizes. These images are divided into 80% and 20%, which are used for training and testing, respectively. The result of the investigation through the experiment is that the proposed deep-learning architecture could achieve an accuracy rate of 95.32%, a precision rate of 94.63%, a specificity of 94.84%, and a high sensitivity of 96.23% using the LUNA16 images. The result shows an improvement compared to a reported accuracy of 93.6% from a previous study using the same dataset.

Helmet and Number Plate Detection

An important use of computer technology in recent years has been the automatic helmet recognition of motorcy clists in real-time surveillance film. Deep learning methods are becoming more and more popular as a result, especially for object detection and classification. Nevertheless, a number ofissues, including limited resolution, inadequate lighting, adverse weather, and occlusion, restrict the accuracy of current models in identifying motorcycle helmets. A unique method that makes use of the Faster R-CNN model has been put out to addressthese issues. Using the input image as the starting point, this method first trains the Region Proposal Network (RPN), and then it uses the RPN weights to train the Faster RCNN model. The goal of this method is to increase helmet detection accuracy in live surveillance footage. This method’s experimental results have demonstrated encouraging results, with a 95% accuracy rate in identifying motorcycle helmets in live surveillance footage.This illustrates the promise of deep learning approaches in the field of automatic helmet detection for motorcyclists in real-time surveillance film, as well as the efficacy of the suggested strategy in overcoming the issues encountered by current models.

Vehicle Detection and Tracking Techniques Based on Deep Learning in Road Traffic Surveillance

This article discusses the application of deep learning in vehicle detection and tracking technology, elaborating on the basic concepts of deep learning and its advantages in vehicle target detection. Deep learning models such as Convolutional Neural Networks (CNNs) overcome the reliance on manual feature engineering by automatically learning image features. The article focuses on two deep learning detection frameworks, Faster R-CNN and YOLO. The former combines region proposal networks with region classification networks to achieve end-to-end optimization, while the latter transforms the detection task into a regression problem, enabling real-time detection in a single forward pass. Regarding vehicle tracking, the article explores the challenges of multi-object tracking such as occlusion, cross-movement, and the tracking requirements of different vehicle types. Deep learning applications in this field, such as the DeepSORT and Tracktor algorithms, combine CNNs, RNNs, and traditional tracking methods to achieve feature learning, historical state modeling, and probabilistic reasoning. Performance evaluation is discussed in terms of metrics like IoU, precision, recall, and F1 Score, comparing and analyzing the performance of different algorithms in vehicle detection and tracking tasks. Lastly, the article discusses the balance between real-time and accuracy in deep learning-based vehicle detection and tracking technology in road traffic monitoring, as well as its significant role in traffic accident warning and management.

MACN: A cascade defect detection for complex background based on mixture attention mechanism

AbstractDefect detection in complex background is a critical issue. To address this issue, this paper proposes the mixture attention mechanism cascade network, in which the new channel attention network is linked with the spatial attention network to create an effective mixed attention network that takes advantage of their respective advantages, adaptively suppresses background noise, and highlights defect features. To ensure the efficiency and effectiveness of effective mixed attention network, the new channel attention network splices the output features of the global average pooling layer and the global maximum pooling layer and then sends the spliced features into a shared network, which is a one‐dimensional convolutional network, and uses cross‐channel interaction for fusion. Furthermore, in order to provide more discriminative feature representation, the authors extract the intermediate features of the region proposal network and input them into effective mixed attention network. Finally, the cascade head is used to refine the predicted bounding box to achieve high‐quality defect location. To demonstrate the superiority and usefulness of this method, it is compared to the latest method using widely used PCB and NEU data sets. A large number of trials demonstrate that this strategy outperforms other methods for detecting defects in complicated backgrounds.

Real-Time Object Detection Using a Lightweight Two-Stage Detection Network with Efficient Data Representation

In this paper, we introduce a novel fast object detection framework, designed to meet the needs of real-time applications such as autonomous driving and robot navigation. Traditional processing methods often trade-off between accuracy and processing speed. To address this issue, we propose a hybrid data representation method that combines the computational efficiency of voxelization with the detail capture capability of direct data processing to optimize overall performance. Our detection framework comprises two main components: a Rapid Region Proposal Network (RPN) and a Refinement Detection Network (RefinerNet). The RPN is used to generate high-quality candidate regions, while the RefinerNet performs detailed analysis on these regions to improve detection accuracy. Additionally, we have implemented a variety of network optimization techniques, including lightweight network layers, network pruning, and model quantization, to increase processing speed and reduce computational resource consumption. Extensive testing on the KITTI and the NEXET datasets has proven the effectiveness of our method in enhancing the accuracy of object detection and real-time processing speed. The experimental results show that, compared to existing technologies, our method performs exceptionally well across multiple evaluation metrics, especially in meeting the stringent requirements of real-time applications in terms of processing speed.

Enhancing lung cancer diagnosis with data fusion and mobile edge computing using DenseNet and CNN

The recent advancements in automated lung cancer diagnosis through the application of Convolutional Neural Networks (CNN) on Computed Tomography (CT) scans have marked a significant leap in medical imaging and diagnostics. The precision of these CNN-based classifiers in detecting and analyzing lung cancer symptoms has opened new avenues in early detection and treatment planning. However, despite these technological strides, there are critical areas that require further exploration and development. In this landscape, computer-aided diagnostic systems and artificial intelligence, particularly deep learning methods like the region proposal network, the dual path network, and local binary patterns, have become pivotal. However, these methods face challenges such as limited interpretability, data variability handling issues, and insufficient generalization. Addressing these challenges is key to enhancing early detection and accurate diagnosis, fundamental for effective treatment planning and improving patient outcomes. This study introduces an advanced approach that combines a Convolutional Neural Network (CNN) with DenseNet, leveraging data fusion and mobile edge computing for lung cancer identification and classification. The integration of data fusion techniques enables the system to amalgamate information from multiple sources, enhancing the robustness and accuracy of the model. Mobile edge computing facilitates faster processing and analysis of CT scan images by bringing computational resources closer to the data source, crucial for real-time applications. The images undergo preprocessing, including resizing and rescaling, to optimize feature extraction. The DenseNet-CNN model, strengthened by data fusion and edge computing capabilities, excels in extracting and learning features from these CT scans, effectively distinguishing between healthy and cancerous lung tissues. The classification categories include Normal, Benign, and Malignant, with the latter further sub-categorized into adenocarcinoma, squamous cell carcinoma, and large cell carcinoma. In controlled experiments, this approach outperformed existing state-of-the-art methods, achieving an impressive accuracy of 99%. This indicates its potential as a powerful tool in the early detection and classification of lung cancer, a significant advancement in medical imaging and diagnostic technology.

MACHINE LEARNING FOR AUTOMATIC EXTRACTION OF WATER BODIES USING SENTINEL-2 IMAGERY

Context. Given the aggravation of environmental and water problems, there is a need to improve automated methods for extracting and monitoring water bodies in urban ecosystems. The problem of efficient and automated extraction of water bodies is becoming relevant given the large amount of data obtained from satellite systems. The object of study is water bodies that are automatically extracted from Sentinel-2 optical satellite images using machine learning methods.
 Objective. The goal of the work is to improve the efficiency of the process of extracting the boundaries of water bodies on digital optical satellite images by using machine learning methods.
 Method. The paper proposes an automated information technology for delineating the boundaries of water bodies on Sentinel-2 digital optical satellite images. The process includes eight stages, starting with data download and using topographic maps to obtain basic information about the study area. Then, the process involved data pre-processing, which included calibrating the images, removing atmospheric noise, and enhancing contrast. Next, the EfficientNet-B0 architecture is applied to identify water features, facilitating optimal network width scaling, depth, and image resolution. ResNet blocks compress and expand channels. It allows for optimal connectivity of large-scale and multi-channel links across layers. After that, the Regional Proposal Network defines regions of interest (ROI), and ROI alignment ensures data homogeneity. The Fully connected layer helps in segmenting the regions, and the Fully connected network creates binary masks for accurate identification of water bodies. The final step of the method is to analyze spatial and temporal changes in the images to identify differences, changes, and trends that may indicate specific phenomena or events. This approach allows automating and accurately identifying water features on satellite images using machine learning.
 Results. The implementation of the proposed technology is development through Python software development. An assessment of the technology’s accuracy, conducted through a comparative analysis with existing methods, such as water indices and K-means, confirms a high level of accuracy in the period from 2017 to 2023 (up to 98%). The Kappa coefficient, which considers the degree of consistency between the actual and predicted classification, confirms the stability and reliability of our approach, reaching a value of 0.96.
 Conclusions. The experiments confirm the effectiveness of the proposed automated information technology and allow us to recommend it for use in studies of changes in coastal areas, decision-making in the field of coastal resource management, and land use. Prospects for further research may include new methods that seasonal changes and provide robustness in the selection and mapping of water surfaces.

SiamATTRPN: Enhance Visual Tracking With Channel and Spatial Attention

Visual tracking is an important research topic in the field of computer vision. The current Siamese tracker based on the region proposal network (SiamRPN) has achieved promising tracking results in terms of efficiency and performance. However, through our empirical study, we have observed that deep features learned by SiamRPN are of substandard quality, as the salient regions within the deep features fail to correspond accurately with meaningful objects. To address this limitation, we propose an approach to enhance the quality of the learned deep features through the incorporation of an attention mechanism. Attention mechanisms have been shown to be effective in distinguishing similar objects, as they suppress background objects while highlighting target information that is most relevant. As a result, a new tracking method with channel and spatial attention termed SiamATTRPN is explored. To verify the effectiveness of SiamATTRPN, experiments on benchmark datasets demonstrate that our proposed tracker outperforms the baseline tracker significantly.

Detection and Classification of Tumors in Brain based on the Location by using MRI

Abstract: Brain tumor is a serious disease occurring in human being. Medical treatment process mainly depends on tumor types and its location. The final decision of neuro specialist and radiologist for the tumor diagnosis mainly depend on evaluation of MRI (Magnetic Resonance Imaging) images. To overcome this, Faster R-CNN deep learning algorithm was proposed for detecting the tumor and marking the area of their occurrence with Region Proposal Network (RPN). The selected MR image dataset consists of three primary brain tumors namely glioma, meningioma and pituitary. The proposed algorithm uses VGG-19 architecture as a base layer for both the region proposal network and the classifier network.

Debiased Novel Category Discovering and Localization

In recent years, object detection in deep learning has experienced rapid development. However, most existing object detection models perform well only on closed-set datasets, ignoring a large number of potential objects whose categories are not defined in the training set. These objects are often identified as background or incorrectly classified as pre-defined categories by the detectors. In this paper, we focus on the challenging problem of Novel Class Discovery and Localization (NCDL), aiming to train detectors that can detect the categories present in the training data, while also actively discover, localize, and cluster new categories. We analyze existing NCDL methods and identify the core issue: object detectors tend to be biased towards seen objects, and this leads to the neglect of unseen targets. To address this issue, we first propose an Debiased Region Mining (DRM) approach that combines class-agnostic Region Proposal Network (RPN) and class-aware RPN in a complementary manner. Additionally, we suggest to improve the representation network through semi-supervised contrastive learning by leveraging unlabeled data. Finally, we adopt a simple and efficient mini-batch K-means clustering method for novel class discovery. We conduct extensive experiments on the NCDL benchmark, and the results demonstrate that the proposed DRM approach significantly outperforms previous methods, establishing a new state-of-the-art.