Depthwise Separable Convolution Research Articles

A 92 fps and 2.56 mJ/Frame Computing-In-Memory-Based Human Pose Estimation Accelerator With Resource-Efficient Macro for Mobile Devices

Human pose estimation (HPE) is a promising solution for accurately understanding the state and context of human actions in virtual reality (VR). A high frame rate with low-power HPE processing is required for a realistic user interaction experience in battery-limited mobile devices. The proposed HPE accelerator is a computing-in-memory (CIM) based accelerator that computes depth-wise separable convolution (DWSC) of a lightweight HPE network. Three key features contribute to a resource-efficient CIM accelerator: 1) Dual-mode CIM computes DWSC with a reconfigurable homogenous architecture, resulting in 2.68 times higher throughput than previous analog CIMs. 2) Effective layeraware unrolling performs bit-parallel computation on dualmode CIM with fewer ADC operations, achieving 46 times higher throughput than before. 3) Adaptive fused inter-macro balancing improves latency balance in layer fusion execution, leading to a 57.0 % higher frame rate than before. The proposed HPE accelerator is implemented in 28nm CMOS technology. It achieves higher computation resource utilization and operates HPE with a low energy-delay product of 27.6 uJ7s in mobile VR devices.

PG-YOLO: An efficient detection algorithm for pomegranate before fruit thinning

The detection of pomegranate fruitlets is of great significance for fruit thinning and yield estimation. However, the researches on the detection of pomegranate fruitlets are rarely considered in recent years. In addition, the immature fruit detection algorithms often have a large amount of calculation, which require high computing memory and computing resources of the equipment. Based on the mentioned issue, we propose a lightweight detection method based on the improved You Only Look Once Version 8(YOLOv8) to realize the automatic detection of pomegranate fruitlets (PG-YOLO). Firstly, the lightweight network Shufflenetv2 is used to reconstruct the backbone of YOLOv8s. Secondly, the Depthwise separable convolution is used to replace the standard convolution of Neck layer. Then, the multi-head self-attention (MHSA) is used to enhance the positioning ability and feature extraction ability of the model before the backbone network outputs features. Finally, the self-built pomegranate fruitlet dataset is used to train and test the PG-YOLO model. It is shown that, the Mean Average Precision(mAP) of pomegranate fruitlets detected by PG-YOLO model is 93.4%. The optimal weight size of PG-YOLO is only 2.2 Mb, which is 89.9% less than the original YOLOv8s. In addition, the detection speed of the PG-YOLO model is 74.1% higher than the original YOLOv8s. The results demonstrate that the study significantly reduces the complexity of the model and provides a research basis for further application in mobile devices. The self-built PG-YOLO-Dataset used in this paper has been published on github: https://github.com/LforikC/PG-YOLO-Dataset.

Mist–fog-assisted real-time emotion recognition using deep transfer learning framework for smart city 4.0

Face emotion recognition (FER) has found several applications across numerous industries. One such application area lies in the field of human–computer interaction, where it can improve user experiences by allowing systems to adjust based on users’ emotional states by leveraging the capabilities of the Internet of Things (IoT) to facilitate more responsive interactions. The increasing demand for precise and immediate emotion analysis across various applications drives the need for FER utilizing deep transfer learning (DTL). In this study, a novel Mist-Fog framework for real-time FER is introduced, which employs a DTL algorithm called Mobile-Net for quick and easy image classification on mobile and embedded devices. It addresses the challenges of operating deep neural networks (DNN) under limits on power and processing resources. By utilizing depth-wise separable convolutions, the MobileNet algorithm reduces computational complexity while maintaining reasonable accuracy on the FER-2013 dataset. A Fog-based architecture, in which a mist layer comprising edge devices like Raspberry-Pi (RPi), an ARM-based single-board computer, was considered for a smart city scenario. Further, the M/M/1/K queuing system was considered, which follows the First-Come-First-Serve (FCFS) scheduling criteria and a Markov Modulated Poisson Process (MMPP) arrival rate to evaluate the system’s performance. The effectiveness of this novel approach towards industrial computer vision paradigms is assessed pertaining to different performance criteria.

Gait-based identification using wearable multimodal sensing and attention neural networks

Advances in sensor technology have sparked a wave of innovation and progress in the domain of wearable devices. However, ensuring the security of these devices remains a critical concern for their users. This study proposes a novel identification framework that employs custom multimodal sensing insoles and deep learning techniques to contribute to the data security of wearable devices in a user-transparent manner. The insole incorporates an inertial sensing unit and ten force sensors to capture kinematic and kinetic information. For this implementation, we propose a lightweight multiclass classification deep model based on depthwise separable convolution and an enhanced attention mechanism, which operates over multiple signal channels. In designing the recognition model, we focused on two pivotal factors: (1) achieving an optimal balance between computational complexity and recognition accuracy, which is essential for models that operate on devices with limited computational power, and (2) introducing an enhanced attention mechanism that ensures high recognition accuracy without the stringent requirement for precise temporal alignment between sensor channels, which can be technically challenging to attain. We conducted extensive experiments to validate the proposed approach, utilizing a walking dataset collected from 82 young subjects. The results indicate that when using data from the multimodal sensing insole, a high recognition accuracy of 97.96 % for person identification is achieved, with a notably low model parameter count of 22,462. The results support the feasibility and effectiveness of sensing footwear-based gait identification. The proposed approach holds significant potential for enhancing user authentication methods across a broad range of scenarios while ensuring transparency and user-friendliness.

An efficient deep neural network for automatic classification of acute intracranial hemorrhages in brain CT scans

BackgroundRecent advancements in deep learning models have demonstrated their potential in the field of medical imaging, achieving remarkable performance surpassing human capabilities in tasks such as classification and segmentation. However, these modern state-of-the-art network architectures often demand substantial computational resources, which limits their practical application in resource-constrained settings. This study aims to propose an efficient diagnostic deep learning model specifically designed for the classification of intracranial hemorrhage in brain CT scans. MethodOur proposed model utilizes a combination of depthwise separable convolutions and a multi-receptive field mechanism to achieve a trade-off between performance and computational efficiency. The model was trained on RSNA datasets and validated on CQ500 dataset and PhysioNet dataset. ResultThrough a comprehensive comparison with state-of-the-art models, our model achieves an average AUROC score of 0.952 on RSNA datasets and exhibits robust generalization capabilities, comparable to SE-ResNeXt, across other open datasets. Furthermore, the parameter count of our model is just 3 % of that of MobileNet V3. ConclusionThis study presents a diagnostic deep-learning model that is optimized for classifying intracranial hemorrhages in brain CT scans. The efficient characteristics make our proposed model highly promising for broader applications in medical settings.

Real-Time Navigation Roads: Lightweight and Efficient Convolutional Neural Network (LE-CNN) for Arabic Traffic Sign Recognition in Intelligent Transportation Systems (ITS)

Smart cities are now embracing the new frontier of urban living, with advanced technology being used to enhance the quality of life for residents. Many of these cities have developed transportation systems that improve efficiency and sustainability, as well as quality. Integrating cutting-edge transportation technology and data-driven solutions improves safety, reduces environmental impact, optimizes traffic flow during peak hours, and reduces congestion. Intelligent transportation systems consist of many systems, one of which is traffic sign detection. This type of system utilizes many advanced techniques and technologies, such as machine learning and computer vision techniques. A variety of traffic signs, such as yield signs, stop signs, speed limits, and pedestrian crossings, are among those that the traffic sign detection system is trained to recognize and interpret. Ensuring accurate and robust traffic sign recognition is paramount for the safe deployment of self-driving cars in diverse and challenging environments like the Arab world. However, existing methods often face many challenges, such as variability in the appearance of signs, real-time processing, occlusions that can block signs, low-quality images, and others. This paper introduces an advanced Lightweight and Efficient Convolutional Neural Network (LE-CNN) architecture specifically designed for accurate and real-time Arabic traffic sign classification. The proposed LE-CNN architecture leverages the efficacy of depth-wise separable convolutions and channel pruning to achieve significant performance improvements in both speed and accuracy compared to existing models. An extensive evaluation of the LE-CNN on the Arabic traffic sign dataset that was carried out demonstrates an impressive accuracy of 96.5% while maintaining superior performance with a remarkably low inference time of 1.65 s, crucial for real-time applications in self-driving cars. It achieves high accuracy with low false positive and false negative rates, demonstrating its potential for real-world applications like autonomous driving and advanced driver-assistance systems.

An accurate lightweight algorithm for Bearings Fault Diagnosis based on DPW ATTCNN model

Rapid and high-precision fault diagnosis of bearings is an essential safeguard for monitoring the safe operation of machinery. Numerous scholars have proposed bearing fault diagnosis models that achieve excellent identification accuracy. However, owing to the elevated cost, some of fault diagnosis models are too complicated to be deployed in practice. This paper proposes a lightweight model based on Depthwise separable convolution (DPW) and Efficient Channel Attention (ECA) called DPW ATTCNN model. The DPW ATTCNN model first adopts DPW to reduce the parameters of the model. Meanwhile, the ECA attention mechanism is utilized to learn crucial information between neural network channels, which improves the accuracy of bearing fault diagnosis. The DPW ATTCNN model is improved based on the Adaptive Batch Normalization (AdaBN) algorithm, which has excellent domain adaptive capability. DPW ATTCNN model achieves excellent performance using a small number of parameters. The experimental results show that the DPW ATTCNN model can achieve 99.58% accuracy in bearing fault diagnosis, which is verified on the public dataset Case Western Reserve University (CWRU) dataset. In addition, the anti-interference evaluation experiment shows that the accuracy of DPW ATTCNN model for bearing fault diagnosis is still higher than 95% even under strong noise interference. In the domain adaptive performance test, the DPW ATTCNN (AdaBN) model achieves an average fault recognition rate of 97.35%, which further verifies the robustness and effectiveness of the proposed DPW ATTCNN algorithm.

Lightweight JPEG image steganalysis using dilated blind-spot network

The primary goal of image steganalysis is to detect subtle steganographic signals hidden within digital images. An essential aspect of digital image steganalysis involves the extraction of meaningful steganographic signal features. However, existing JPEG steganalysis networks have not delved deep enough into the research on extracting steganographic noise signals, especially in the preprocessing stage of the model. In this paper, we propose a novel approach called the Dilated Blind-Spot Steganalysis Network (DBS2Net), specifically designed for JPEG steganalysis. DBS2Net incorporates the dilated blind-spot convolutional network into the preprocessing stage to extract noise residuals, thereby enhancing detection accuracy. The blind-spot network estimates central outputs based on adjacent regions to effectively suppress semantic information and uses dilated convolutions to expand the receptive field. Additionally, we design the transition block to further enhance noise signals by mapping noise residuals extracted by the blind-spot network. In feature extraction, depth-wise separable convolutions are used to minimize the parameter count, while dual-view pooling layers compress features, resulting in improved detection performance. Experimental results validate the efficacy of this approach, and comprehensive analyses of JPEG steganographic noise support its success. In particular, DBS2Net outperforms state-of-the-art steganalysis methods, achieving superior detection accuracy with minimal model complexity.

BAF-Net: bidirectional attention-aware fluid pyramid feature integrated multimodal fusion network for diagnosis and prognosis

Objective. To go beyond the deficiencies of the three conventional multimodal fusion strategies (i.e. input-, feature- and output-level fusion), we propose a bidirectional attention-aware fluid pyramid feature integrated fusion network (BAF-Net) with cross-modal interactions for multimodal medical image diagnosis and prognosis. Approach. BAF-Net is composed of two identical branches to preserve the unimodal features and one bidirectional attention-aware distillation stream to progressively assimilate cross-modal complements and to learn supplementary features in both bottom-up and top-down processes. Fluid pyramid connections were adopted to integrate the hierarchical features at different levels of the network, and channel-wise attention modules were exploited to mitigate cross-modal cross-level incompatibility. Furthermore, depth-wise separable convolution was introduced to fuse the cross-modal cross-level features to alleviate the increase in parameters to a great extent. The generalization abilities of BAF-Net were evaluated in terms of two clinical tasks: (1) an in-house PET-CT dataset with 174 patients for differentiation between lung cancer and pulmonary tuberculosis. (2) A public multicenter PET-CT head and neck cancer dataset with 800 patients from nine centers for overall survival prediction. Main results. On the LC-PTB dataset, improved performance was found in BAF-Net (AUC = 0.7342) compared with input-level fusion model (AUC = 0.6825; p < 0.05), feature-level fusion model (AUC = 0.6968; p = 0.0547), output-level fusion model (AUC = 0.7011; p < 0.05). On the H&N cancer dataset, BAF-Net (C-index = 0.7241) outperformed the input-, feature-, and output-level fusion model, with 2.95%, 3.77%, and 1.52% increments of C-index (p = 0.3336, 0.0479 and 0.2911, respectively). The ablation experiments demonstrated the effectiveness of all the designed modules regarding all the evaluated metrics in both datasets. Significance. Extensive experiments on two datasets demonstrated better performance and robustness of BAF-Net than three conventional fusion strategies and PET or CT unimodal network in terms of diagnosis and prognosis.

Lightweight object detection network for multi‐damage recognition of concrete bridges in complex environments

AbstractTo solve the challenges of low recognition accuracy, slow speed, and weak generalization ability inherent in traditional methods for multi‐damage recognition of concrete bridges, this paper proposed an efficient lightweight damage recognition model, constructed by improving the you only look once v4 (YOLOv4) with MobileNetv3 and fused inverted residual blocks, named YOLOMF. First, a novel lightweight network named MobileNetv3 with fused inverted residual (MobileNetv3‐FusedIR) is constructed as the backbone network for YOLOMF. This is achieved by integrating the fused mobile inverted bottleneck convolution (Fused‐MBConv) into the shallow layers of MobileNetv3. Second, the standard convolution in YOLOv4 is replaced with the depthwise separable convolution, resulting in a reduction in the number of parameters and complexity of the model. Third, the effects of different activation functions on the damage recognition performance of YOLOMF are thoroughly investigated. Finally, to verify the effectiveness of the proposed method in complex environments, a data enhancement library named Imgaug is used to simulate concrete bridge damage images under challenging conditions such as motion blur, fog, rain, snow, noise, and color variations. The results indicate that the YOLOMF shows excellent multi‐damage recognition proficiency for concrete bridges across varying field‐of‐view sizes as well as complex environmental conditions. The detection speed of YOLOMF reaches 85f/s, facilitating effective real‐time multi‐damage detection for concrete bridges under complex environments.

Feature Sparse Choosing VIT Model for Efficient Concrete Crack Segmentation in Portable Crack Measuring Devices

Concrete crack measurement is important for concrete buildings. Deep learning-based segmentation methods have achieved state-of-art results. However, the model size of these models is extremely large which is impossible to use in portable crack measuring devices. To address this problem, a light-weight concrete crack segmentation model based on the Feature Sparse Choosing VIT (LTNet) is proposed by us. In our proposed model, a Feature Sparse Choosing VIT (FSVIT) is used to reduce computational complexity in VIT as well as reducing the number of channels for crack features. In addition, a Feature Channel Selecting Module (FCSM) is proposed by us to reduce channel features as well as suppressing the influence of interfering features. Finally, Depthwise Separable Convolutions are used to substitute traditional convolutions for further reducing computational complexity. As a result, the model size of our LTNet is extremely small. Experimental results show that our LTNet could achieve an accuracy of 0.887, 0.817 and 0.693, and achieve a recall of 0.882, 0.805 and 0.681 on three datasets, respectively, which is 3–8% higher than current mainstream algorithms. However, the model size of our LTNet is only 2 M.

SEHRNet: A lightweight, high‐resolution network for aircraft keypoint detection

AbstractCurrent research on apron conflict detection is often limited to the interaction between the aircraft as a whole and other objects, making it difficult to accomplish targeted identification of vulnerable and high‐cost aircraft components. However, the implementation of detailed aircraft identification is of great significance to enhance the safety of airport surface operations. Based on the excellent performance of High‐Resolution Network (HRNet) in keypoint detection, a lightweight end‐to‐end keypoint detection network, namely Squeeze and Excitation High‐Resolution Network (SEHRNet), is proposed in this paper to solve the problems of HRNet's slower computation and more redundancy. First, the errors arising from coordinate transformations in the coding and decoding process are solved by an improved feature map coding and decoding process. Second, replace the BasicBlock in HRNet with the Depthwise separable convolutions based on the Squeeze‐and‐Excitation Networks, which drastically cuts the computational cost of the network. Third, the improved Bottleneck module is used to further enhance the capability of feature extraction. Experimental results prove that, based on the aircraft keypoint detection dataset, the SEHRNet proposed in this paper shows stronger applicability compared to the current mainstream networks. Compared with the original HRNet, the improved network has higher accuracy, faster speed, and lighter model for aircraft keypoint detection.

A framework for electricity load forecasting based on attention mechanism time series depthwise separable convolutional neural network

Electricity load exhibits daily and weekly cyclical patterns as well as random characteristics. At present, prevailing deep learning models cannot learn electricity load cyclical and stochastic features adequately. This results in insufficient prediction accuracy and the scalability of current methods. To tackle these difficulties, this paper proposes a framework for electrical load prediction based on an Attention Mechanism Time Series Depthwise Separable Convolutional Neural Network (ELPF-ATDSCN). The framework starts by using the Maximum Information Coefficient for exogenous variable selection. It then incorporates a seasonal decomposition algorithm with manual feature engineering to extract the cyclical and stochastic features of the electrical load. Subsequently, the framework employs the ATDSCN to learn the cyclical and stochastic features of the electrical load. In addition, the Bayesian algorithm optimizes model hyperparameters for optimal model performance. Experimental results of point and interval load prediction on datasets from the US and Nordic power markets reveal that the ATDSCN model proposed in this paper enhances load prediction accuracy compared with other models. It can provide more reliable predictions for power system operation and dispatch.

AVD-YOLOv5: a new lightweight network architecture for high-speed aortic valve detection from a new and large echocardiography dataset.

Heart disease detection is currently gaining widespread attention as a means to enhance the accuracy of cardiologists' diagnoses from cardiac images and reduce diagnosis time. Although high-resolution computed tomography (CT) images are typically favored for heart disease detection, the drawbacks of cost and radiation exposure to patients necessitate alternative approaches. In this context, utilizing ultrasound images becomes pivotal to mitigate radiation risks and maintain cost-effectiveness. In this paper, we propose a novel lightweight model, AVD-YOLOv5, designed for automated aortic valve detection on echocardiography images. This model incorporates several enhancements to the YOLOv5 architecture. Notably, the depth-wise separable convolution significantly contributes to the model's lightweight design by reducing the number of parameters while maintaining precision. We have also created a new and larger dataset comprising 260 echocardiography images specifically for aortic valve detection. Experimental results indicate that the precision value of the modified ADV-YOLOv5 model stands at 94.3%, with a recall value of 86.8%. The model also demonstrates a notable 67% reduction in inference time compared to the original YOLOv5 model. Although there is a marginal reduction in precision by 0.94%, the model's efficiency is significantly increased. The proposed system can be used by cardiologists for more efficient and reliable diagnosis.

Multi-attention bidirectional contrastive learning method for unpaired image-to-image translation.

Unpaired image-to-image translation (I2IT) involves establishing an effective mapping between the source and target domains to enable cross-domain image transformation. Previous contrastive learning methods inadequately accounted for the variations in features between two domains and the interrelatedness of elements within the features. Consequently, this can result in challenges encompassing model instability and the blurring of image edge features. To this end, we propose a multi-attention bidirectional contrastive learning method for unpaired I2IT, referred to as MabCUT. We design separate embedding blocks for each domain based on depthwise separable convolutions and train them simultaneously from both the source and target domains. Then we utilize a pixel-level multi-attention extractor to query images from embedding blocks in order to select feature blocks with crucial information, thus preserving essential features from the source domain. To enhance the feature representation capability of the model, we incorporate depthwise separable convolutions for the generator. We conducted comprehensive evaluations using three datasets, demonstrating that our approach enhances the quality of unpaired I2IT while avoiding the issue of mode collapse-related image blurring.

BT-CNN: a balanced binary tree architecture for classification of brain tumour using MRI imaging.

Deep learning is a very important technique in clinical diagnosis and therapy in the present world. Convolutional Neural Network (CNN) is a recent development in deep learning that is used in computer vision. Our medical investigation focuses on the identification of brain tumour. To improve the brain tumour classification performance a Balanced binary Tree CNN (BT-CNN) which is framed in a binary tree-like structure is proposed. It has a two distinct modules-the convolution and the depthwise separable convolution group. The usage of convolution group achieves lower time and higher memory, while the opposite is true for the depthwise separable convolution group. This balanced binarty tree inspired CNN balances both the groups to achieve maximum performance in terms of time and space. The proposed model along with state-of-the-art models like CNN-KNN and models proposed by Musallam et al., Saikat et al., and Amin et al. are experimented on public datasets. Before we feed the data into model the images are pre-processed using CLAHE, denoising, cropping, and scaling. The pre-processed dataset is partitioned into training and testing datasets as per 5 fold cross validation. The proposed model is trained and compared its perforarmance with state-of-the-art models like CNN-KNN and models proposed by Musallam et al., Saikat et al., and Amin et al. The proposed model reported average training accuracy of 99.61% compared to other models. The proposed model achieved 96.06% test accuracy where as other models achieved 68.86%, 85.8%, 86.88%, and 90.41% respectively. Further, the proposed model obtained lowest standard deviation on training and test accuracies across all folds, making it invariable to dataset.

DP-GAN+B: A lightweight generative adversarial network based on depthwise separable convolutions for generating CT volumes

X-rays, commonly used in clinical settings, offer advantages such as low radiation and cost-efficiency. However, their limitation lies in the inability to distinctly visualize overlapping organs. In contrast, Computed Tomography (CT) scans provide a three-dimensional view, overcoming this drawback but at the expense of higher radiation doses and increased costs. Hence, from both the patient's and hospital's standpoints, there is substantial medical and practical value in attempting the reconstruction from two-dimensional X-ray images to three-dimensional CT images. In this paper, we introduce DP-GAN+B as a pioneering approach for transforming two-dimensional frontal and lateral lung X-rays into three-dimensional lung CT volumes. Our method innovatively employs depthwise separable convolutions instead of traditional convolutions and introduces vector and fusion loss for superior performance. Compared to prior models, DP-GAN+B significantly reduces the generator network parameters by 21.104 M and the discriminator network parameters by 10.82 M, resulting in a total reduction of 31.924 M (44.17%). Experimental results demonstrate that our network can effectively generate clinically relevant, high-quality CT images from X-ray data, presenting a promising solution for enhancing diagnostic imaging while mitigating cost and radiation concerns.

Enhanced multimodal emotion recognition in healthcare analytics: A deep learning based model-level fusion approach

Deep learning techniques have drawn considerable interest in emotion recognition due to recent technological developments in healthcare analytics. Automatic patient emotion recognition can assist healthcare analytics by providing feedback to the stakeholders of competent healthcare about the conditions of the patients and their satisfaction levels. In this paper, we propose a novel model-level fusion technique based on deep learning for enhanced emotion recognition from multimodal signals to monitor patients in connected healthcare. The representative visual features from the video signals are extracted through the Depthwise Separable Convolution Neural Network, and the optimized temporal attributes are derived from the multiple physiological data utilizing Bi-directional Long Short-Term Memory. A soft attention method fused the high multimodal features obtained from the two data modalities to retrieve the most significant features by focusing on emotionally salient parts of the features. We exploited two face detection methods, Histogram of Oriented Gradients and Convolutional Neural Network-based face detector (ResNet-34), to observe the effects of facial features on emotion recognition. Lastly, extensive experimental evaluations have been conducted using the widely used Bio Vid Emo DB multimodal dataset to verify the performance of the proposed architecture. Experimental results show that the developed fusion architecture improved the accuracy of emotion recognition from multimodal signals and outperformed the performance of both state-of-the-art techniques and baseline methods.

Offline Handwriting Writer Identification using Depth-wise Separable Convolution with Siamese Network

Offline handwriting writer identification has significant implications for forensic investigations and biometric authentication. Handwriting, as a distinctive biometric trait, provides insights into individual identity. Despite advancements in handcrafted algorithms and deep learning techniques, the persistent challenges related to intra-variability and inter-writer similarity continue to drive research efforts. In this study, we build on well-separated convolution architectures like the Xception architecture, which has proven to be robust in our previous research comparing various deep learning architectures such as MobileNet, EfficientNet, ResNet50, and VGG16, where Xception demonstrated minimal training-validation disparities for writer identification. Expanding on this, we use a model based on similarity or dissimilarity approaches to identify offline writers' handwriting, known as the Siamese Network, that incorporates the Xception architecture. Similarity or dissimilarity measurements are based on the Manhattan or L1 distance between representation vectors of each input pair. We train publicly available IAM and CVL datasets; our approach achieves accuracy rates of 99.81% for IAM and 99.88% for CVL. The model was evaluated using evaluation metrics, which revealed only two error predictions in the IAM dataset, resulting in 99.75% accuracy, and five error predictions for CVL, resulting in 99.57% accuracy. These findings modestly surpass existing achievements, highlighting the potential inherent in our methodology to enhance writer identification accuracy. This study underscores the effectiveness of integrating the Siamese Network with depth-wise separable convolution, emphasizing the practical implications for supporting writer identification in real-world applications.

Potato leaf disease detection with a novel deep learning model based on depthwise separable convolution and transformer networks

Early diagnosis of plant diseases is essential in reducing economic losses for farmers and increasing production efficiency. Therefore, Computer-Aided Diagnosis (CAD) systems supported by artificial intelligence technologies can be developed to help diagnose diseases quickly and accurately by examining the symptoms and signs in plant leaves. In this study, Multi-head Attention Mechanism Depthwise Separable Convolution Inception Reduction Network (MDSCIRNet) architecture, an image-based deep convolutional neural network, is proposed for classifying potato leaf diseases. The main components of the MDSCIRNet architecture are depthwise separable convolution (DSC) and a multi-head attention mechanism. The proposed architecture has been compared with modern algorithms developed with DSC technology, such as Xception, MobileNet, and deep learning algorithms, such as ResNet101, InceptionV3, and EfficientNetB2, to evaluate its performance in the classification process. In addition, hybrid methods developed with the classical machine learning algorithms Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), Adaptive Boosting (AdaBoost), and MDSCIRNet model, integrated deep learning model, hard voting ensemble learning model. Suggested methods such as these were also used in the experimental process. Moreover, techniques such as Contrast Limited Adaptive Histogram Equalization (CLAHE), Enhanced Super-Resolution Generative Adversarial Networks (ESRGAN), and Hypercolumn were used to improve the image quality of the data set. In the experimental process, the MDSCIRNet deep learning architecture achieved 99.24% accuracy in the study using the original dataset. While a 99.11% accuracy rate was achieved with the integrated deep learning model and hard voting ensemble learning model, the highest success rate of 99.33% was performed in the study conducted with the MDSCIRNet + SVM method. This study contributes to developing new and effective strategies in the agricultural industry for the early diagnosis and control of potato plant diseases. Machine learning-based approaches offer the potential to minimize economic losses and increase productivity in production by allowing farmers to intervene early.