Mohammad Hossein Shakoor
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Mohammad Hossein Shakoor has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Reza Boostani
CSE & IT Department, Faculty of Electrical and Computer Engineering, Shiraz University, Shiraz, Iran
6 shared publications - Malihe Sabeti
Department of Computer Engineering, North Tehran Branch, Islamic Azad University, Tehran, Iran
3 shared publications - Abouzar Zareei
Department of Computer Engineering Faculty of Engineering Arak University Arak 38156-8-8349 Iran araku.ac.ir
2 shared publications - Mokhtar Mohammadi
Department of Information Technology, Lebanese French University, Erbil, Iraq
2 shared publications - Adil Hussein Mohammed
Department of Communication and Computer Engineering, Cihan University-Erbil, Kurdistan Region, Iraq
1 shared publication Ali Khodadadi
Department of Computer Engineering, Faculty of Engineering, Arak University, Arak, Iran
1 shared publication
Latest Publications
- Open Access
- 4 August 2026
- Published by Elsevier in Intelligence-Based Medicine
- Vol. 15
Efficient few-shot learning for medical image classification: A practical review of lightweight and resource-aware approaches
- 4 August 2026
- Published by Elsevier in Intelligence-Based Medicine
- Vol. 15
Open AccessEfficient few-shot learning for medical image classification: A practical review of lightweight and resource-aware approaches
- Open Access
- 25 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 82749-82775
A Comprehensive Study on the Evolution of Different Motivations and Architectures of YOLO
YOLO (You Only Look Once) is a popular real-time object detection method that has been widely used in most computer vision applications recently. YOLO is a fast and lightweight deep neural network that can provide the output in a single-stage operation. This paper is a comprehensive survey of more than 200 papers that have been published over the last two years about YOLO applications. This study explains the applications of YOLO and their motivations to change the architecture to enhance their operations. In this research all of the comparison criteria of different YOLO versions have been compared and the YOLO components have been revised comprehensively. The advantages and limitations of each YOLO versions have been discussed and the statistical comparisons of different YOLO versions in different applications have been illustrated. These applications include industrial, medical, agriculture, agronomy, transportation, security, monitoring, remote sensing, and animal-related. These applications are related to the classification, segmentation, detection, recognition, and localization of different objects in the grayscale, colored, X-ray, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), Infrared (IR), Thermal, and synthetic aperture radar (SAR) images. For all of the revised papers the motivation, the improvement and the specifications of that application have been discussed and for some of these applications the results have been shown in some figures. In terms of the motivations of each reviewed paper, some researches replaced the backbone of the original YOLO with a lightweight network to reduce the parameters, FLOPs (floating point operations), and model size. Some of them, proposed an attention mechanism to enhance feature extraction capabilities of backbone. Diffusion mechanism of features and dimension integration of the neck are other innovations in most of the YOLO-based approaches to optimize the detection of small and overlapped objects. Furthermore, some of these researchers improved the loss function and head enhancement of YOLO to boost detection accuracy.
- 25 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 82749-82775
Open AccessA Comprehensive Study on the Evolution of Different Motivations and Architectures of YOLO
YOLO (You Only Look Once) is a popular real-time object detection method that has been widely used in most computer vision applications recently. YOLO is a fast and lightweight deep neural network that can provide the output in a single-stage operation. This paper is a comprehensive survey of more than 200 papers that have been published over the last two years about YOLO applications. This study explains the applications of YOLO and their motivations to change the architecture to enhance their operations. In this research all of the comparison criteria of different YOLO versions have been compared and the YOLO components have been revised comprehensively. The advantages and limitations of each YOLO versions have been discussed and the statistical comparisons of different YOLO versions in different applications have been illustrated. These applications include industrial, medical, agriculture, agronomy, transportation, security, monitoring, remote sensing, and animal-related. These applications are related to the classification, segmentation, detection, recognition, and localization of different objects in the grayscale, colored, X-ray, Computed Tomography (CT), Magnetic Resonance Imaging (MRI), Infrared (IR), Thermal, and synthetic aperture radar (SAR) images. For all of the revised papers the motivation, the improvement and the specifications of that application have been discussed and for some of these applications the results have been shown in some figures. In terms of the motivations of each reviewed paper, some researches replaced the backbone of the original YOLO with a lightweight network to reduce the parameters, FLOPs (floating point operations), and model size. Some of them, proposed an attention mechanism to enhance feature extraction capabilities of backbone. Diffusion mechanism of features and dimension integration of the neck are other innovations in most of the YOLO-based approaches to optimize the detection of small and overlapped objects. Furthermore, some of these researchers improved the loss function and head enhancement of YOLO to boost detection accuracy.
- Open Access
- 1 January 2026
- Published by Institution of Engineering and Technology (IET) in IET Biometrics
- Vol. 2026 (1)
Compact Fingerprint Identification With a Parameter‐Free LBP Front End: A Source‐Level Controlled Ablation Study of Representations and Classifier Heads
This study investigates a parameter‐free local binary pattern (LBP) front end for compact closed‐set fingerprint identification under a source‐impression‐level leakage‐controlled protocol. The official FVC2000 DB4_B set provides 80 original impressions from 10 fingers. For each of five seeded partitions, five impressions per finger are used for training, one for validation and two for testing. Ten deterministic views are generated only after source partitioning, yielding frozen 500/100/200‐image arrays; the 20 held‐out source impressions, rather than their 200 correlated views, define the primary evaluation unit. Twenty CNN configurations form 10 matched GAP/flatten pairs spanning raw inputs, scalar and bit‐plane LBP, multiscale encoding and backbone depth. All receive 200 epochs of online translation augmentation plus five unaugmented epochs; the checkpoint with minimum validation loss is used for primary evaluation. The deeper normalized‐raw Flatten model achieves 91.0% ± 6.5% source‐level accuracy, whereas a standard unnormalized‐raw GAP model achieves 84.0% ± 4.2% with 19,466 parameters. GAP substantially reduces classifier capacity but does not consistently improve accuracy; effects depend on input encoding and network depth. Statistical comparisons across five repeated splits are interpreted descriptively owing to their limited resolution.
- 1 January 2026
- Published by Institution of Engineering and Technology (IET) in IET Biometrics
- Vol. 2026 (1)
Open AccessCompact Fingerprint Identification With a Parameter‐Free LBP Front End: A Source‐Level Controlled Ablation Study of Representations and Classifier Heads
This study investigates a parameter‐free local binary pattern (LBP) front end for compact closed‐set fingerprint identification under a source‐impression‐level leakage‐controlled protocol. The official FVC2000 DB4_B set provides 80 original impressions from 10 fingers. For each of five seeded partitions, five impressions per finger are used for training, one for validation and two for testing. Ten deterministic views are generated only after source partitioning, yielding frozen 500/100/200‐image arrays; the 20 held‐out source impressions, rather than their 200 correlated views, define the primary evaluation unit. Twenty CNN configurations form 10 matched GAP/flatten pairs spanning raw inputs, scalar and bit‐plane LBP, multiscale encoding and backbone depth. All receive 200 epochs of online translation augmentation plus five unaugmented epochs; the checkpoint with minimum validation loss is used for primary evaluation. The deeper normalized‐raw Flatten model achieves 91.0% ± 6.5% source‐level accuracy, whereas a standard unnormalized‐raw GAP model achieves 84.0% ± 4.2% with 19,466 parameters. GAP substantially reduces classifier capacity but does not consistently improve accuracy; effects depend on input encoding and network depth. Statistical comparisons across five repeated splits are interpreted descriptively owing to their limited resolution.
- Open Access
- 4 September 2025
- Published by National Academy of Sciences of Tajikistan in InfoScience Trends
- Vol. 2 (9) , 39-51
Attention Mechanisms in U-Net Variants for Brain MRI Segmentation: A Narrative Review of Spatial, Channel, and Transformer-Based Approaches
Accurately delineating brain tumors in magnetic resonance imaging (MRI) scans is a critical step in clinical diagnostics, radiotherapy planning, and evaluating therapeutic outcomes. However, several obstacles, including low image contrast, irregular tumor boundaries, and intensity inhomogeneities, often complicate this task. Deep learning methodologies, especially the U-Net architecture, have become prominent tools for automating medical image segmentation. Yet, the traditional U-Net model is limited by its inability to selectively highlight important features or to incorporate extensive contextual information, which constrains its performance in complex imaging scenarios. To overcome these drawbacks, researchers have introduced various attention mechanisms into the U-Net framework, thereby improving its capability to focus dynamically on pertinent spatial and semantic features. This review offers a critical analysis of four notable U-Net variants enhanced with attention modules: Attention U-Net, which integrates gated spatial attention; SE U-Net, employing channel-specific feature recalibration; CBAM-UNet, which combines channel and spatial attention in sequence; and TransUNet, which incorporates Transformer-based self-attention to model long-range dependencies. We comprehensively examine the architectural designs, strategies for integrating attention, segmentation accuracy, interpretability, and computational efficiency of these models. Furthermore, we address prevailing challenges such as limited generalizability, sensitivity to noise, and issues related to clinical adoption. Prospective advancements are also discussed, including hybrid attention strategies, the development of explainable AI, and the integration of multimodal data. This review provides a focused synthesis of key attention-augmented U-Net models for brain tumor segmentation, highlighting their architectural innovations, comparative performance, and practical considerations for clinical use.
- 4 September 2025
- Published by National Academy of Sciences of Tajikistan in InfoScience Trends
- Vol. 2 (9) , 39-51
Open AccessAttention Mechanisms in U-Net Variants for Brain MRI Segmentation: A Narrative Review of Spatial, Channel, and Transformer-Based Approaches
Accurately delineating brain tumors in magnetic resonance imaging (MRI) scans is a critical step in clinical diagnostics, radiotherapy planning, and evaluating therapeutic outcomes. However, several obstacles, including low image contrast, irregular tumor boundaries, and intensity inhomogeneities, often complicate this task. Deep learning methodologies, especially the U-Net architecture, have become prominent tools for automating medical image segmentation. Yet, the traditional U-Net model is limited by its inability to selectively highlight important features or to incorporate extensive contextual information, which constrains its performance in complex imaging scenarios. To overcome these drawbacks, researchers have introduced various attention mechanisms into the U-Net framework, thereby improving its capability to focus dynamically on pertinent spatial and semantic features. This review offers a critical analysis of four notable U-Net variants enhanced with attention modules: Attention U-Net, which integrates gated spatial attention; SE U-Net, employing channel-specific feature recalibration; CBAM-UNet, which combines channel and spatial attention in sequence; and TransUNet, which incorporates Transformer-based self-attention to model long-range dependencies. We comprehensively examine the architectural designs, strategies for integrating attention, segmentation accuracy, interpretability, and computational efficiency of these models. Furthermore, we address prevailing challenges such as limited generalizability, sensitivity to noise, and issues related to clinical adoption. Prospective advancements are also discussed, including hybrid attention strategies, the development of explainable AI, and the integration of multimodal data. This review provides a focused synthesis of key attention-augmented U-Net models for brain tumor segmentation, highlighting their architectural innovations, comparative performance, and practical considerations for clinical use.
- Open Access
- 3 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 127690-127720
Noise-Robust Local Ternary Pattern Center for Noisy Texture Classification
Local Binary Pattern (LBP) has been widely used for texture analysis in machine vision and image processing. However, most of these descriptors provide a high number of features that are sensitive to noise and rotation. In this paper, a version of LTP is proposed that is named Local Ternary Pattern Center (LTPC). This descriptor provides noise-robust and low feature numbers with a better classification rate than LTP and advanced LBP versions for noisy textures. The proposed method has three steps. First a criterion is proposed that estimates the quality of the texture. According to the estimated quality of the texture, an average kernel is applied to the noisy texture. The more level of noise the more number of applying average kernel must be convolved to each texture. Convolution is a time-consuming operation, so here, a fast average method is proposed that can do this operation around 2 to 3 times faster than the traditional convolution method. In the second step LTPC is used to extract the features of filtered textures. The third innovation of this paper is related to the mapping of the features. Most of the advanced LBP versions used the riu2 mapping method to decrease the number of features and make them rotation-invariant. In this research, a novel mapping method is proposed. It has all advantages of the riu2 and it provides features that they are more robust to noise than riu2. The implementations on different datasets such as Outex suits (TC10, TC12(t), TC12(h), TC13), CUReT, UIUC, USPTex, Virus, RSMAS, KTHTIPS2, UMD and ORL indicate that the proposed method can provide very high discriminative and noise-robust features. The higher the level of noise the better classification accuracy can be obtained by the proposed method compared to some state-of-the-art LBP versions. Furthermore the classification rate of proposed method is better than some deep neural network results. Whereas, in term of speed the local descriptors provide very shorter operation time than most of the deep learning methods.
- 3 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 127690-127720
Open AccessNoise-Robust Local Ternary Pattern Center for Noisy Texture Classification
Local Binary Pattern (LBP) has been widely used for texture analysis in machine vision and image processing. However, most of these descriptors provide a high number of features that are sensitive to noise and rotation. In this paper, a version of LTP is proposed that is named Local Ternary Pattern Center (LTPC). This descriptor provides noise-robust and low feature numbers with a better classification rate than LTP and advanced LBP versions for noisy textures. The proposed method has three steps. First a criterion is proposed that estimates the quality of the texture. According to the estimated quality of the texture, an average kernel is applied to the noisy texture. The more level of noise the more number of applying average kernel must be convolved to each texture. Convolution is a time-consuming operation, so here, a fast average method is proposed that can do this operation around 2 to 3 times faster than the traditional convolution method. In the second step LTPC is used to extract the features of filtered textures. The third innovation of this paper is related to the mapping of the features. Most of the advanced LBP versions used the riu2 mapping method to decrease the number of features and make them rotation-invariant. In this research, a novel mapping method is proposed. It has all advantages of the riu2 and it provides features that they are more robust to noise than riu2. The implementations on different datasets such as Outex suits (TC10, TC12(t), TC12(h), TC13), CUReT, UIUC, USPTex, Virus, RSMAS, KTHTIPS2, UMD and ORL indicate that the proposed method can provide very high discriminative and noise-robust features. The higher the level of noise the better classification accuracy can be obtained by the proposed method compared to some state-of-the-art LBP versions. Furthermore the classification rate of proposed method is better than some deep neural network results. Whereas, in term of speed the local descriptors provide very shorter operation time than most of the deep learning methods.
- 21 April 2025
- Published by Springer Nature in International Journal of Information Technology
- p. 1-16
Repeated game theory kernel for support vector machine
- 21 April 2025
- Published by Springer Nature in International Journal of Information Technology
- p. 1-16
Repeated game theory kernel for support vector machine
- Open Access
- 24 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 22473-22488
Eliminating Network Depth: Genetic Algorithm for Parameter Optimization in CNNs
Recent advances in Convolutional Neural Networks (CNNs) have significantly enhanced image classification performance. However, CNNs often require large numbers of parameters, leading to increased computational complexity, prolonged training times, and substantial resource demands. Achieving higher classification accuracy typically involves deepening network architectures, which further exacerbates these challenges. This paper proposes a novel method based on a genetic algorithm to optimize parameter selection, enabling the construction of CNNs that achieve superior accuracy with fewer parameters. By focusing on parameters with the most significant impact on performance, the method reduces the need for deeper networks, thereby minimizing computational costs. Experimental results demonstrate that the proposed algorithm outperforms its counterparts. For instance, the generated CNN achieves an accuracy improvement of 0.75 percentage points over ResNet-110 while using 84% fewer parameters. These findings highlight the method’s potential to balance efficiency and accuracy, making it a promising solution for resource-constrained applications.
- 24 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 22473-22488
Open AccessEliminating Network Depth: Genetic Algorithm for Parameter Optimization in CNNs
Recent advances in Convolutional Neural Networks (CNNs) have significantly enhanced image classification performance. However, CNNs often require large numbers of parameters, leading to increased computational complexity, prolonged training times, and substantial resource demands. Achieving higher classification accuracy typically involves deepening network architectures, which further exacerbates these challenges. This paper proposes a novel method based on a genetic algorithm to optimize parameter selection, enabling the construction of CNNs that achieve superior accuracy with fewer parameters. By focusing on parameters with the most significant impact on performance, the method reduces the need for deeper networks, thereby minimizing computational costs. Experimental results demonstrate that the proposed algorithm outperforms its counterparts. For instance, the generated CNN achieves an accuracy improvement of 0.75 percentage points over ResNet-110 while using 84% fewer parameters. These findings highlight the method’s potential to balance efficiency and accuracy, making it a promising solution for resource-constrained applications.
- Open Access
- 1 December 2024
- Published by Apex Publishing in Journal of Green Energy Research and Innovation
- Vol. 1 (4) , 64-85
A Survey on Renewable and New Sources of Energies for Electricity Power Production and Its Challenges
Common sources of electricity production, such as fossil fuels, create a large amount of pollution; furthermore, these sources will run out in the future. Therefore, the use of new sources and renewable energies are very vital for electricity production. This paper provides a comprehensive study to explore and analyze new and renewable electrical energy sources for the production of electricity. The study assesses the awareness, perceptions, and preferences of individuals regarding innovative technologies and sources that have the potential to reshape the landscape of electrical power generation. In this research, various types of novel and renewable sources for electricity production are explained and the challenges, advantages, and disadvantages of each of these methods are discussed.
- 1 December 2024
- Published by Apex Publishing in Journal of Green Energy Research and Innovation
- Vol. 1 (4) , 64-85
Open AccessA Survey on Renewable and New Sources of Energies for Electricity Power Production and Its Challenges
Common sources of electricity production, such as fossil fuels, create a large amount of pollution; furthermore, these sources will run out in the future. Therefore, the use of new sources and renewable energies are very vital for electricity production. This paper provides a comprehensive study to explore and analyze new and renewable electrical energy sources for the production of electricity. The study assesses the awareness, perceptions, and preferences of individuals regarding innovative technologies and sources that have the potential to reshape the landscape of electrical power generation. In this research, various types of novel and renewable sources for electricity production are explained and the challenges, advantages, and disadvantages of each of these methods are discussed.
- Open Access
- 1 September 2024
- Published by Apex Publishing in Journal of Green Energy Research and Innovation
- Vol. 1 (3) , 96-109
A Survey of Different Methods for Miner Detection and Challenges of Them in Power Industries
Cryptocurrency mining requires high consumption power. In recent years, because of the increase in the price of cryptocurrencies and due to the cheap price of electricity in Iran, mining Bitcoin and other cryptocurrencies has been very profitable. Some miners are legally engaged in cryptocurrency mining, but many miners do it illegally and without permission. Since cryptocurrency mining is an operation that consumes a lot of electricity, it is one of the reasons for the lack of electricity, especially in the summer season, and it has caused power outages and financial losses to various industries. In this paper, different methods of detection of illegal mining are reviewed. In this research, by collecting the small number of researches done in the world, a comprehensive study of this issue has been tried. The methods of identifying miners and related consumers are divided into several different categories. Some of them are based-on data mining and mostly to identify consumers who use the output of the converter for the miner device. The second category is related to people who illegally get their electricity from behind the meter or unmetered branches, which are much more difficult to identify than the first type. These methods will be mentioned more in this article. Finally, some suggestions are provided for better identification of these consumers. Furthermore, at the end of the paper, some renewable and new sources of electrical power are discussed for using as an electricity power for miners instead of traditional fossil fuel and gas planes.
- 1 September 2024
- Published by Apex Publishing in Journal of Green Energy Research and Innovation
- Vol. 1 (3) , 96-109
Open AccessA Survey of Different Methods for Miner Detection and Challenges of Them in Power Industries
Cryptocurrency mining requires high consumption power. In recent years, because of the increase in the price of cryptocurrencies and due to the cheap price of electricity in Iran, mining Bitcoin and other cryptocurrencies has been very profitable. Some miners are legally engaged in cryptocurrency mining, but many miners do it illegally and without permission. Since cryptocurrency mining is an operation that consumes a lot of electricity, it is one of the reasons for the lack of electricity, especially in the summer season, and it has caused power outages and financial losses to various industries. In this paper, different methods of detection of illegal mining are reviewed. In this research, by collecting the small number of researches done in the world, a comprehensive study of this issue has been tried. The methods of identifying miners and related consumers are divided into several different categories. Some of them are based-on data mining and mostly to identify consumers who use the output of the converter for the miner device. The second category is related to people who illegally get their electricity from behind the meter or unmetered branches, which are much more difficult to identify than the first type. These methods will be mentioned more in this article. Finally, some suggestions are provided for better identification of these consumers. Furthermore, at the end of the paper, some renewable and new sources of electrical power are discussed for using as an electricity power for miners instead of traditional fossil fuel and gas planes.
- Open Access
- 17 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 91673-91695
Weighted Constraint Feature Selection of Local Descriptor for Texture Image Classification
There are several statistical descriptors for feature extraction from texture images. Local binary pattern is one of the most popular descriptor for revealing the underlying structure of a texture. Recently several variants of local binary descriptor have been proposed. The completed local binary pattern is an efficient version that can provide discriminant features and consequently providing a high classification rate. It finely characterizes a texture by fusing three histograms of features. Fusing histograms is applied by jointing the histograms and it increases the feature number significantly; therefore, in this paper a weighted constraint feature selection approach is proposed to select a very small number of features without any degradation in classification accuracy. It significantly enhances the classification rate by using a very low number of informative features. The proposed feature selection approach is a filter-based feature selection. It employed a weighted constraint score for each feature. After ranking the features, a threshold estimation method is proposed to select the most discriminant features. For a better comparison, a wide range of different datasets is used as benchmark to assess the compared methods. Implementations on Outex, UIUC, CUReT, MeasTex, Brodatz, Virus, Coral Reef, and ORL face datasets indicate that the proposed method can provide high classification accuracy without any learning step just by selecting a few features of descriptor.
- 17 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 91673-91695
Open AccessWeighted Constraint Feature Selection of Local Descriptor for Texture Image Classification
There are several statistical descriptors for feature extraction from texture images. Local binary pattern is one of the most popular descriptor for revealing the underlying structure of a texture. Recently several variants of local binary descriptor have been proposed. The completed local binary pattern is an efficient version that can provide discriminant features and consequently providing a high classification rate. It finely characterizes a texture by fusing three histograms of features. Fusing histograms is applied by jointing the histograms and it increases the feature number significantly; therefore, in this paper a weighted constraint feature selection approach is proposed to select a very small number of features without any degradation in classification accuracy. It significantly enhances the classification rate by using a very low number of informative features. The proposed feature selection approach is a filter-based feature selection. It employed a weighted constraint score for each feature. After ranking the features, a threshold estimation method is proposed to select the most discriminant features. For a better comparison, a wide range of different datasets is used as benchmark to assess the compared methods. Implementations on Outex, UIUC, CUReT, MeasTex, Brodatz, Virus, Coral Reef, and ORL face datasets indicate that the proposed method can provide high classification accuracy without any learning step just by selecting a few features of descriptor.