Backbone Architecture Research Articles

Data Augmentation with Generative Adversarial Network for Solar Panel Segmentation from Remote Sensing Images

With the popularity of solar energy in the electricity market, demand rises for data such as precise locations of solar panels for efficient energy planning and management. However, these data are not easily accessible; information such as precise locations sometimes does not exist. Furthermore, existing datasets for training semantic segmentation models of photovoltaic (PV) installations are limited, and their annotation is time-consuming and labor-intensive. Therefore, for additional remote sensing (RS) data creation, the pix2pix generative adversarial network (GAN) is used, enriching the original resampled training data of varying ground sampling distances (GSDs) without compromising their integrity. Experiments with the DeepLabV3 model, ResNet-50 backbone, and pix2pix GAN architecture were conducted to discover the advantage of using GAN-based data augmentations for a more accurate RS imagery segmentation model. The result is a fine-tuned solar panel semantic segmentation model, trained using transfer learning and an optimal amount—60% of GAN-generated RS imagery for additional training data. The findings demonstrate the benefits of using GAN-generated images as additional training data, addressing the issue of limited datasets, and increasing IoU and F1 metrics by 2% and 1.46%, respectively, compared with classic augmentations.

Uncertainty-aware Cross-Entropy for Semantic Segmentation

Abstract. Deep neural networks have shown exceptional performance in various tasks, but their lack of robustness, reliability, and tendency to be overconfident pose challenges for their deployment in safety-critical applications like autonomous driving. In this regard, quantifying the uncertainty inherent to a model’s prediction is a promising endeavour to address these shortcomings. In this work, we present a novel Uncertainty-aware Cross-Entropy loss (U-CE) that incorporates dynamic predictive uncertainties into the training process by pixel-wise weighting of the well-known cross-entropy loss (CE). Through extensive experimentation, we demonstrate the superiority of U-CE over regular CE training on two benchmark datasets, Cityscapes and ACDC, using two common backbone architectures, ResNet-18 and ResNet-101. With U-CE, we manage to train models that not only improve their segmentation performance but also provide meaningful uncertainties after training. Consequently, we contribute to the development of more robust and reliable segmentation models, ultimately advancing the state-of-the-art in safety-critical applications and beyond.

OD-YOLO: Robust Small Object Detection Model in Remote Sensing Image with a Novel Multi-Scale Feature Fusion.

As remote sensing technology has advanced, the use of satellites and similar technologies has become increasingly prevalent in daily life. Now, it plays a crucial role in hydrology, agriculture, and geography. Nevertheless, because of the distinct qualities of remote sensing, including expansive scenes and small, densely packed targets, there are many challenges in detecting remote sensing objects. Those challenges lead to insufficient accuracy in remote sensing object detection. Consequently, developing a new model is essential to enhance the identification capabilities for objects in remote sensing imagery. To solve these constraints, we have designed the OD-YOLO approach that uses multi-scale feature fusion to improve the performance of the YOLOv8n model in small target detection. Firstly, traditional convolutions have poor recognition capabilities for certain geometric shapes. Therefore, in this paper, we introduce the Detection Refinement Module (DRmodule) into the backbone architecture. This module utilizes Deformable Convolutional Networks and the Hybrid Attention Transformer to strengthen the model's capability for feature extraction from geometric shapes and blurred objects effectively. Meanwhile, based on the Feature Pyramid Network of YOLO, at the head of the model framework, this paper enhances the detection capability by introducing a Dynamic Head to strengthen the fusion of different scales features in the feature pyramid. Additionally, to address the issue of detecting small objects in remote sensing images, this paper specifically designs the OIoU loss function to finely describe the difference between the detection box and the true box, further enhancing model performance. Experiments on the VisDrone dataset show that OD-YOLO surpasses the compared models by at least 5.2% in mAP50 and 4.4% in mAP75, and experiments on the Foggy Cityscapes dataset demonstrated that OD-YOLO improved mAP by 6.5%, demonstrating outstanding results in tasks related to remote sensing images and adverse weather object detection. This work not only advances the research in remote sensing image analysis, but also provides effective technical support for the practical deployment of future remote sensing applications.

Vehicle detection in varied weather conditions using enhanced deep YOLO with complex wavelet

Traffic congestion is prevalent in many major and medium-sized cities throughout different countries in contemporary society. In traffic images, various multi-sized vehicles are tightly clustered together and obstructed from one another. Identifying vehicles in such instances is crucial for urban traffic surveillance, safety monitoring, and legal concerns but it also presents major challenges. The remarkable detection accuracy and efficiency of deep learning-based systems have led to their recent and extensive use in vehicle identification. There are significant advanced YOLO models with different backbone architectures and frameworks developed for vehicle detection. Yet, the performance of YOLO variants are facing the challenges of handling false detection against occluded and densely sophisticated scenarios. The proposed model is developed to address such types of limitations, for example; dynamic illumination, noisy images, and scale sensitivity to improve the vehicle detection rate in different traffic scenarios and varying weather conditions. The proposed study employs an improved YOLOv4 to identify moving vehicles in different lighting conditions including daylight, cloudy, rainy, and night. For hybridization, three techniques are utilized such as the Multiscale Retinex, Dual tree complex wavelet transform (DTCWT), and Pulse Coupled Neural Networks (PCNN). The DTCWT is employed for multiscale decomposition and to denoise the complex high frequency subband information, then the denoised subbands are reconstructed into a denoised image. The Multiscale retinex is utilized to reduce the halo artifacts on high-contrast edges and maintain the balance with dynamic range compression and color reproduction. The synchronizing pulse burst property of PCNN is used to detect the isolated noisy pixels and modify the detected noisy pixels. From the results it is worth noting that the developed model surpasses state-of-the-art methods in sunny, night, cloudy, and rainy modes. The proposed method using the DTCWT technique can detect the vehicles with mAP of 91.09% and 35FPS.

Deep Learning-based Lightweight Models for Bactrocera Oleae Detection

The detection of Bactrocera oleae (BO), an invasive pest that poses a serious threat to olive tree cultivation, is crucial for ensuring the quality and quantity of olive oil production. This study explores the application of deep learning with lightweight models for BO detection in olive trees. The DIRT Dataset, consisting of 841 images obtained from traps, was used for training and evaluation. Different combinations of Faster R-CNN and SSD models with various backbone architectures, including EfficientNet B0, MobileNet v3, ResNet-18, and ResNet-50, were assessed. The combination of Faster R-CNN with ResNet-18 achieves comparable results to Faster R-CNN with ResNet-50 but with lower latency and a reduced parameter count. These findings highlight the potential of deep learning-based object detection models for BO detection in olive trees, offering efficient and accurate identification of infestations.

Deep-learning model for background parenchymal enhancement classification in contrast-enhanced mammography

Background. Breast background parenchymal enhancement (BPE) is correlated with the risk of breast cancer. BPE level is currently assessed by radiologists in contrast-enhanced mammography (CEM) using 4 classes: minimal, mild, moderate and marked, as described in breast imaging reporting and data system (BI-RADS). However, BPE classification remains subject to intra- and inter-reader variability. Fully automated methods to assess BPE level have already been developed in breast contrast-enhanced MRI (CE-MRI) and have been shown to provide accurate and repeatable BPE level classification. However, to our knowledge, no BPE level classification tool is available in the literature for CEM. Materials and methods. A BPE level classification tool based on deep learning has been trained and optimized on 7012 CEM image pairs (low-energy and recombined images) and evaluated on a dataset of 1013 image pairs. The impact of image resolution, backbone architecture and loss function were analyzed, as well as the influence of lesion presence and type on BPE assessment. The evaluation of the model performance was conducted using different metrics including 4-class balanced accuracy and mean absolute error. The results of the optimized model for a binary classification: minimal/mild versus moderate/marked, were also investigated. Results. The optimized model achieved a 4-class balanced accuracy of 71.5% (95% CI: 71.2–71.9) with 98.8% of classification errors between adjacent classes. For binary classification, the accuracy reached 93.0%. A slight decrease in model accuracy is observed in the presence of lesions, but it is not statistically significant, suggesting that our model is robust to the presence of lesions in the image for a classification task. Visual assessment also confirms that the model is more affected by non-mass enhancements than by mass-like enhancements. Conclusion. The proposed BPE classification tool for CEM achieves similar results than what is published in the literature for CE-MRI.

Efficient multi-task progressive learning for semantic segmentation and disparity estimation

Scene understanding is an important area in robotics and autonomous driving. To accomplish these tasks, the 3D structures in the scene have to be inferred to know what the objects and their locations are. To this end, semantic segmentation and disparity estimation networks are typically used, but running them individually is inefficient since they require high-performance resources. A possible solution is to learn both tasks together using a multi-task approach. Some current methods address this problem by learning semantic segmentation and monocular depth together. However, monocular depth estimation from single images is an ill-posed problem. A better solution is to estimate the disparity between two stereo images and take advantage of this additional information to improve the segmentation. This work proposes an efficient multi-task method that jointly learns disparity and semantic segmentation. Employing a Siamese backbone architecture for multi-scale feature extraction, the method integrates specialized branches for disparity estimation and coarse and refined segmentations, leveraging progressive task-specific feature sharing and attention mechanisms to enhance accuracy for solving both tasks concurrently. The proposal achieves state-of-the-art results for joint segmentation and disparity estimation on three distinct datasets: Cityscapes, TrimBot2020 Garden, and S-ROSeS, using only 1/3 of the parameters of previous approaches.

Feature Maps Need More Attention: A Spatial-Channel Mutual Attention-Guided Transformer Network for Face Super-Resolution

Recently, transformer-based face super-resolution (FSR) approaches have achieved promising success in restoring degraded facial details due to their high capability for capturing both local and global dependencies. However, while existing methods focus on introducing sophisticated structures, they neglect the potential feature map information, limiting FSR performance. To circumvent this problem, we carefully design a pair of guiding blocks to dig for possible feature map information to enhance features before feeding them to transformer blocks. Relying on the guiding blocks, we propose a spatial-channel mutual attention-guided transformer network for FSR, for which the backbone architecture is a multi-scale connected encoder–decoder. Specifically, we devise a novel Spatial-Channel Mutual Attention-guided Transformer Module (SCATM), which is composed of a Spatial-Channel Mutual Attention Guiding Block (SCAGB) and a Channel-wise Multi-head Transformer Block (CMTB). SCATM on the top layer (SCATM-T) aims to promote both local facial details and global facial structures, while SCATM on the bottom layer (SCATM-B) seeks to optimize the encoded features. Considering that different scale features are complementary, we further develop a Multi-scale Feature Fusion Module (MFFM), which fuses features from different scales for better restoration performance. Quantitative and qualitative experimental results on various datasets indicate that the proposed method outperforms other state-of-the-art FSR methods.

Optimizing facial feature extraction and localization using YOLOv5: An empirical analysis of backbone architectures with data augmentation for precise facial region detection

Optimizing facial feature extraction and localization using YOLOv5: An empirical analysis of backbone architectures with data augmentation for precise facial region detection

Memory Guided Transformer With Spatio-Semantic Visual Extractor for Medical Report Generation.

Medicalimaging-based report writing for effective diagnosis in radiology is time-consuming and can be error-prone by inexperienced radiologists. Automatic reporting helps radiologists avoid missed diagnoses and saves valuable time. Recently, transformer-based medical report generation has become prominent in capturing long-term dependencies of sequential data with its attention mechanism. Nevertheless, input features obtained from traditional visual extractor of conventional transformers do not capture spatial and semantic information of an image. So, the transformer is unable to capture fine-grained details and may not produce detailed descriptive reports of radiology images. Therefore, we propose a spatio-semantic visual extractor (SSVE) to capture multi-scale spatial and semantic information from radiology images. Here, we incorporate two types of networks in ResNet 101 backbone architecture, i.e. (i) deformable network at the intermediate layer of ResNet 101 that utilizes deformable convolutions in order to obtain spatially invariant features, and (ii) semantic network at the final layer of backbone architecture which uses dilated convolutions to extract rich multi-scale semantic information. Further, these network representations are fused to encode fine-grained details of radiology images. The performance of our proposed model outperforms existing works on two radiology report datasets, i.e., IU X-ray and MIMIC-CXR.

Image fusion using Y-net-based extractor and global-local discriminator

Although some deep learning-based image fusion approaches have realized promising results, how to extract information-rich features from different source images while preserving them in the fused image with less distortions remains challenging issue that needs to be addressed. Here, we propose a well worked-out GAN-based scheme with multi-scale feature extractor and global-local discriminator for infrared and visible image fusion. We use Y-Net as the backbone architecture to design the generator network, and introduce the residual dense block (RDblock) to yield more realistic fused images for infrared and visible images by learning discriminative multi-scale representations that are closer to the essence of different modal images. During feature reconstruction, the cross-modality shortcuts with contextual attention (CMSCA) are employed to selectively aggregate features at different scales and different levels to construct information-rich fused images with better visual effect. To ameliorate the information content of the fused image, we not only constrain the structure and contrast information using structural similarity index, but also evaluate the intensity and gradient similarities at both feature and image levels. Two global-local discriminators that combine global GAN with PatchGAN as a unified architecture help to dig for finer differences between the generated image and reference images, which force the generator to learn both the local radiation information and pervasive global details in two source images. It is worth mentioning that image fusion is achieved during confrontation without fusion rules. Lots of assessment tests demonstrate that the reported fusion scheme achieves superior performance against state-of-the-art works in meaningful information preservation.

An optimised YOLOv4 deep learning model for efficient malarial cell detection in thin blood smear images.

Malaria is a serious public health concern worldwide. Early and accurate diagnosis is essential for controlling the disease's spread and avoiding severe health complications. Manual examination of blood smear samples by skilled technicians is a time-consuming aspect of the conventional malaria diagnosis toolbox. Malaria persists in many parts of the world, emphasising the urgent need for sophisticated and automated diagnostic instruments to expedite the identification of infected cells, thereby facilitating timely treatment and reducing the risk of disease transmission. This study aims to introduce a more lightweight and quicker model-but with improved accuracy-for diagnosing malaria using a YOLOv4 (You Only Look Once v. 4) deep learning object detector. The YOLOv4 model is modified using direct layer pruning and backbone replacement. The primary objective of layer pruning is the removal and individual analysis of residual blocks within the C3, C4 and C5 (C3-C5) Res-block bodies of the backbone architecture's C3-C5 Res-block bodies. The CSP-DarkNet53 backbone is simultaneously replaced for enhanced feature extraction with a shallower ResNet50 network. The performance metrics of the models are compared and analysed. The modified models outperform the original YOLOv4 model. The YOLOv4-RC3_4 model with residual blocks pruned from the C3 and C4 Res-block body achieves the highest mean accuracy precision (mAP) of 90.70%. This mAP is > 9% higher than that of the original model, saving approximately 22% of the billion floating point operations (B-FLOPS) and 23MB in size. The findings indicate that the YOLOv4-RC3_4 model also performs better, with an increase of 9.27% in detecting the infected cells upon pruning the redundant layers from the C3 Res-block bodies of the CSP-DarkeNet53 backbone. The results of this study highlight the use of the YOLOv4 model for detecting infected red blood cells. Pruning the residual blocks from the Res-block bodies helps to determine which Res-block bodies contribute the most and least, respectively, to the model's performance. Our method has the potential to revolutionise malaria diagnosis and pave the way for novel deep learning-based bioinformatics solutions. Developing an effective and automated process for diagnosing malaria will considerably contribute to global efforts to combat this debilitating disease. We have shown that removing undesirable residual blocks can reduce the size of the model and its computational complexity without compromising its precision.

A medical image segmentation method for rectal tumors based on multi-scale feature retention and multiple attention mechanisms.

With the continuous development of deep learning algorithms in the field of medical images, models for medical image processing based on convolutional neural networks have made great progress. Since medical images of rectal tumors are characterized by specific morphological features and complex edges that differ from natural images, achieving good segmentation results often requires a higher level of enrichment through the utilization of semantic features. The efficiency of feature extraction and utilization has been improved to some extent through enhanced hardware arithmetic and deeper networks in most models. However, problems still exist with detail loss and difficulty in feature extraction, arising from the extraction of high-level semantic features in deep networks. In this work, a novel medical image segmentation model has been proposed for Magnetic Resonance Imaging (MRI) image segmentation of rectal tumors. The model constructs a backbone architecture based on the idea of jump-connected feature fusion and solves the problems of detail feature loss and low segmentation accuracy using three novel modules: Multi-scale Feature Retention (MFR), Multi-branch Cross-channel Attention (MCA), and Coordinate Attention (CA). Compared with existing methods, our proposed model is able to segment the tumor region more effectively, achieving 97.4% and 94.9% in Dice and mIoU metrics, respectively, exhibiting excellent segmentation performance and computational speed. Our proposed model has improved the accuracy of both lesion region and tumor edge segmentation. In particular, the determination of the lesion region can help doctors identify the tumor location in clinical diagnosis, and the accurate segmentation of the tumor edge can assist doctors in judging the necessity and feasibility of surgery.

DeepX-Ray: A Comparative Study of Deep Learning-Based Classification and Segmentation Techniques for Automated Detection and Diagnosis of COVID-19 from Chest X-ray Images

The relentless spread of the SARS-CoV-2 virus, causing COVID-19, has underscored the urgent need for efficient early detection and diagnosis methods to mitigate its impact. While traditional techniques like RT-PCR are valuable, they often suffer from time-consuming processes. In this study, DeepX-Ray is presented, a comprehensive investigation into deep learning-based classification and segmentation approaches for the automated detection of COVID-19 from chest X-ray images. Specifically, the authors focus on constructing a custom convolutional neural network (CNN) model to distinguish between COVID-19 and standard X-ray images and benchmark its performance against established models. For segmentation tasks, the effectiveness of various backbone architectures, including ResNet34, ResNet101, DenseNet201, and ResNet50 within the UNet model framework, is explored, with ResNet50 exhibiting superior performance. Furthermore, a novel dataset comprising 7657 images sourced from three publicly available authentic datasets is introduced, and the labelMe tool is employed by the authors to generate ground truth mask datasets for 4137 images to facilitate segmentation of infected lung areas. The authors' custom CNN model achieves an outstanding classification accuracy of 100%, while the segmentation approach attains a mean Intersection over Union (IoU) score of 96.19%. These results underscore the efficacy of the proposed model in enabling early automatic detection and diagnosis of COVID-19, particularly in resource-constrained and remote settings where establishing traditional laboratories may be impractical. This research significantly advances medical imaging techniques for combating the COVID-19 pandemic.

YOLOX-SwinT algorithm improves the accuracy of AO/OTA classification of intertrochanteric fractures by orthopedic trauma surgeons

PurposeIntertrochanteric fracture (ITF) classification is crucial for surgical decision-making. However, orthopedic trauma surgeons have shown lower accuracy in ITF classification than expected. The objective of this study was to utilize an artificial intelligence (AI) method to improve the accuracy of ITF classification. MethodsWe trained a network called YOLOX-SwinT, which is based on the You Only Look Once X (YOLOX) object detection network with Swin Transformer (SwinT) as the backbone architecture, using 762 radiographic ITF examinations as the training set. Subsequently, we recruited 5 senior orthopedic trauma surgeons (SOTS) and 5 junior orthopedic trauma surgeons (JOTS) to classify the 85 original images in the test set, as well as the images with the prediction results of the network model in sequence. Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS) 20.0 (IBM Corp., Armonk, NY, USA) to compare the differences among the SOTS, JOTS, SOTS + AI, JOTS + AI, SOTS + JOTS, and SOTS + JOTS + AI groups. All images were classified according to the AO/OTA 2018 classification system by 2 experienced trauma surgeons and verified by another expert in this field. Based on the actual clinical needs, after discussion, we integrated 8 subgroups into 5 new subgroups, and the dataset was divided into training, validation, and test sets by the ratio of 8:1:1. ResultsThe mean average precision at the intersection over union (IoU) of 0.5 (mAP50) for subgroup detection reached 90.29%. The classification accuracy values of SOTS, JOTS, SOTS + AI, and JOTS + AI groups were 56.24% ± 4.02%, 35.29% ± 18.07%, 79.53% ± 7.14%, and 71.53% ± 5.22%, respectively. The paired t-test results showed that the difference between the SOTS and SOTS + AI groups was statistically significant, as well as the difference between the JOTS and JOTS + AI groups, and the SOTS + JOTS and SOTS + JOTS + AI groups. Moreover, the difference between the SOTS + JOTS and SOTS + JOTS + AI groups in each subgroup was statistically significant, with all p < 0.05. The independent samples t-test results showed that the difference between the SOTS and JOTS groups was statistically significant, while the difference between the SOTS + AI and JOTS + AI groups was not statistically significant. With the assistance of AI, the subgroup classification accuracy of both SOTS and JOTS was significantly improved, and JOTS achieved the same level as SOTS. ConclusionIn conclusion, the YOLOX-SwinT network algorithm enhances the accuracy of AO/OTA subgroups classification of ITF by orthopedic trauma surgeons.

Neuromorphic Event Signal-Driven Network for Video De-raining

Convolutional neural networks-based video de-raining methods commonly rely on dense intensity frames captured by CMOS sensors. However, the limited temporal resolution of these sensors hinders the capture of dynamic rainfall information, limiting further improvement in de-raining performance. This study aims to overcome this issue by incorporating the neuromorphic event signal into the video de-raining to enhance the dynamic information perception. Specifically, we first utilize the dynamic information from the event signal as prior knowledge, and integrate it into existing de-raining objectives to better constrain the solution space. We then design an optimization algorithm to solve the objective, and construct a de-raining network with CNNs as the backbone architecture using a modular strategy to mimic the optimization process. To further explore the temporal correlation of the event signal, we incorporate a spiking self-attention module into our network. By leveraging the low latency and high temporal resolution of the event signal, along with the spatial and temporal representation capabilities of convolutional and spiking neural networks, our model captures more accurate dynamic information and significantly improves de-raining performance. For example, our network achieves a 1.24dB improvement on the SynHeavy25 dataset compared to the previous state-of-the-art method, while utilizing only 39% of the parameters.

EAN: An Efficient Attention Module Guided by Normalization for Deep Neural Networks

Deep neural networks (DNNs) have achieved remarkable success in various fields, and two powerful techniques, feature normalization and attention mechanisms, have been widely used to enhance model performance. However, they are usually considered as two separate approaches or combined in a simplistic manner. In this paper, we investigate the intrinsic relationship between feature normalization and attention mechanisms and propose an Efficient Attention module guided by Normalization, dubbed EAN. Instead of using costly fully-connected layers for attention learning, EAN leverages the strengths of feature normalization and incorporates an Attention Generation (AG) unit to re-calibrate features. The proposed AG unit exploits the normalization component as a measure of the importance of distinct features and generates an attention mask using GroupNorm, L2 Norm, and Adaptation operations. By employing a grouping, AG unit and aggregation strategy, EAN is established, offering a unified module that harnesses the advantages of both normalization and attention, while maintaining minimal computational overhead. Furthermore, EAN serves as a plug-and-play module that can be seamlessly integrated with classic backbone architectures. Extensive quantitative evaluations on various visual tasks demonstrate that EAN achieves highly competitive performance compared to the current state-of-the-art attention methods while sustaining lower model complexity.

Enhancing Generalization in Few-Shot Learning for Detecting Unknown Adversarial Examples

Deep neural networks, particularly convolutional neural networks, are vulnerable to adversarial examples, undermining their reliability in visual recognition tasks. Adversarial example detection is a crucial defense mechanism against such attacks but often relies on empirical observations and specialized metrics, posing challenges in terms of data efficiency, generalization to unknown attacks, and scalability to high-resolution datasets like ImageNet. To address these issues, we propose a prototypical network-based method using a deep residual network as the backbone architecture. This approach is capable of extracting discriminative features of adversarial and normal examples from various known adversarial examples by constructing few-shot adversarial detection tasks. Then the optimal mapping matrix is computed using the Sinkhorn algorithm from optimal transport theory, and the class centers are iteratively updated, enabling the detection of unknown adversarial examples across scenarios. Experimental results show that the proposed approach outperforms existing methods in the cross-adversary benchmark and achieves enhanced generalization on a subset of ImageNet in detecting both new adversarial attacks and adaptive white-box attacks. The proposed approach offers a promising solution for improving the safety of deep neural networks in practical applications.

ResNet50-Boosted UNet for Improved Liver Segmentation Accuracy

Segmentation of the liver from abdominal CT images is difficult due to changes in form, density, and the presence of malignancies. This research describes a novel strategy to improve segmentation accuracy that uses UNet as a foundation architecture and ResNet50 as a backbone architecture. This integrated design automates feature selection and spatial awareness, overcoming limitations in previous models. Experimental evaluations using the LiTS dataset show higher performance. Specifically, using the LiTS dataset, our algorithm achieves a remarkable foreground accuracy of 99.81% in liver segmentation. These results outperform existing approaches, demonstrating UNet and ResNet50's potential as valuable tools for precise liver segmentation in clinical situations. The suggested system shows promise for application in diverse medical imaging tasks other than liver segmentation, demonstrating its versatility and effectiveness in enhancing machine-assisted medical diagnostics and decision-making processes.

Backbone search for object detection for applications in intrusion warning systems

&lt;span&gt;In this work, we propose a novel backbone search method for object detection for applications in intrusion warning systems. The goal is to find a compact model for use in embedded thermal imaging cameras widely used in intrusion warning systems. The proposed method is based on faster region-based convolutional neural network (Faster R-CNN) because it can detect small objects. Inspired by EfficientNet, the sought-after backbone architecture is obtained by finding the most suitable width scale for the base backbone (ResNet50). The evaluation metrics are mean average precision (mAP), number of parameters, and number of multiply–accumulate operations (MACs). The experimental results showed that the proposed method is effective in building a lightweight neural network for the task of object detection. The obtained model can keep the predefined mAP while minimizing the number of parameters and computational resources. All experiments are executed elaborately on the person detection in intrusion warning systems (PDIWS) dataset. &lt;/span&gt;