Ali Ahmadinia
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Ali Ahmadinia has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Ahmed Saeed
School of Engineering and Built Environment, Glasgow Caledonian University, Glasgow, UK
5 shared publications - Abbas Javed
Department of Computer Engineering, COMSATS University Islamabad, Lahore Campus, Lahore, Pakistan
3 shared publications Hadi Larijani
School of Engineering and Built Environment, Glasgow Caledonian University, Glasgow, United Kingdom
3 shared publicationsEmmanuel A. Castillo
Department of Computer Science and Information Systems, California State University San Marcos, San Marcos, California 92069, USA
2 shared publicationsJared Macshane
California State University,Department of Computer Science and Information Systems,San Marcos,CA,USA
2 shared publicationsKambiz Hamadani
California State Univ. (United States)
2 shared publications
Latest Publications
- Open Access
- 22 June 2026
- Published by Institution of Engineering and Technology (IET) in Energy Conversion and Economics
Towards practical oscillation detection in wind farms: Comparative study of AI models, novel metrics, and edge implementations
Fast oscillations observed in wind farms and other renewable energy systems interfaced with power grids must be addressed promptly to prevent devastating shutdown and damage. Existing approaches mainly target high‐magnitude oscillations using data‐intensive convolutional neural networks (CNNs), leading to delayed detection and overlooking mild oscillations that can still cause long‐term degradation. This paper proposes a dual detection framework that rapidly identifies high‐magnitude oscillations using time‐domain analysis while detecting mild oscillations through FFT‐based frequency‐domain feature extraction. The framework leverages hyperdimensional computing (HDC) and machine learning models, including one‐class support vector machine (SVM), 1D‐CNN, 2D‐CNN, and a one‐class autoencoder. To enhance performance evaluation, we introduce true positive confidence (TPC) and false positive confidence (FPC) metrics. Unlike prior work limited to server‐based evaluation, we deploy HDC on edge platforms, including a Jetson Nano graphics processing unit (GPU), tensor processing unit (TPU), and field programmable gate array (FPGA). Experimental results on real‐world wind farm data and Simulink‐generated datasets show that HDC outperforms all baseline models, achieving at least 99% confidence in both TPC and FPC for high‐magnitude and mild oscillations. Notably, HDC operates up to 2100 times faster than 1D‐CNN on GPU and 6800 times faster on FPGA, while being four times smaller in model size, demonstrating that FPGA‐based HDC is the minimum suitable hardware for reliable real‐time deployment in wind farms.
- 22 June 2026
- Published by Institution of Engineering and Technology (IET) in Energy Conversion and Economics
Open AccessTowards practical oscillation detection in wind farms: Comparative study of AI models, novel metrics, and edge implementations
Fast oscillations observed in wind farms and other renewable energy systems interfaced with power grids must be addressed promptly to prevent devastating shutdown and damage. Existing approaches mainly target high‐magnitude oscillations using data‐intensive convolutional neural networks (CNNs), leading to delayed detection and overlooking mild oscillations that can still cause long‐term degradation. This paper proposes a dual detection framework that rapidly identifies high‐magnitude oscillations using time‐domain analysis while detecting mild oscillations through FFT‐based frequency‐domain feature extraction. The framework leverages hyperdimensional computing (HDC) and machine learning models, including one‐class support vector machine (SVM), 1D‐CNN, 2D‐CNN, and a one‐class autoencoder. To enhance performance evaluation, we introduce true positive confidence (TPC) and false positive confidence (FPC) metrics. Unlike prior work limited to server‐based evaluation, we deploy HDC on edge platforms, including a Jetson Nano graphics processing unit (GPU), tensor processing unit (TPU), and field programmable gate array (FPGA). Experimental results on real‐world wind farm data and Simulink‐generated datasets show that HDC outperforms all baseline models, achieving at least 99% confidence in both TPC and FPC for high‐magnitude and mild oscillations. Notably, HDC operates up to 2100 times faster than 1D‐CNN on GPU and 6800 times faster on FPGA, while being four times smaller in model size, demonstrating that FPGA‐based HDC is the minimum suitable hardware for reliable real‐time deployment in wind farms.
- 7 June 2023
- Published by SPIE-Intl Soc Optical Eng
- Vol. 12701, 43
Tracking objects using QR codes and deep learning
Despite recent advances in deep learning, object detection and tracking still require considerable manual and computational effort. First, we need to collect and create a database of hundreds or thousands of images of the target objects. Next we must annotate or curate the images to indicate the presence and position of the target objects within those images. Finally, we must train a CNN (convolution neural network) model to detect and locate the target objects in new images. This training is usually computationally intensive, consists of thousands of epochs, and can take tens of hours for each target object. Even after the model training in completed, there is still a chance of failure if the real-time tracking and object detection phases lack sufficient accuracy, precision, and/or speed for many important applications. Here we present a system and approach which minimizes the computational expense of the various steps in the training and real-time tracking process outlined above of for applications in the development of mixed-reality science laboratory experiences by using non-intrusive object-encoding 2D QR codes that are mounted directly onto the surfaces of the lab tools to be tracked. This system can start detecting and tracking it immediately and eliminates the laborious process of acquiring and annotating a new training dataset for every new lab tool to be tracked.
- 7 June 2023
- Published by SPIE-Intl Soc Optical Eng
- Vol. 12701, 43
Tracking objects using QR codes and deep learning
Despite recent advances in deep learning, object detection and tracking still require considerable manual and computational effort. First, we need to collect and create a database of hundreds or thousands of images of the target objects. Next we must annotate or curate the images to indicate the presence and position of the target objects within those images. Finally, we must train a CNN (convolution neural network) model to detect and locate the target objects in new images. This training is usually computationally intensive, consists of thousands of epochs, and can take tens of hours for each target object. Even after the model training in completed, there is still a chance of failure if the real-time tracking and object detection phases lack sufficient accuracy, precision, and/or speed for many important applications. Here we present a system and approach which minimizes the computational expense of the various steps in the training and real-time tracking process outlined above of for applications in the development of mixed-reality science laboratory experiences by using non-intrusive object-encoding 2D QR codes that are mounted directly onto the surfaces of the lab tools to be tracked. This system can start detecting and tracking it immediately and eliminates the laborious process of acquiring and annotating a new training dataset for every new lab tool to be tracked.
- 27 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 37-42
AI Assisted Trail Map Generation based on Public GPS Data
Hiking trail maps are typically created manually by survey, a time-consuming process. This process is expensive and must be repeated over time to improve accuracy. This paper proposed an inexpensive, automatic, and accurate trail network generation method from anonymous public GPS data utilizing a growing self-organizing map (GSOM). This technique does not rely on sequential GPS traces to learn network topology, unlike other approaches. Tuning several hyper-parameters can adjust this process for datasets and networks with unique characteristics. Reconstruction and adaption are also possible based on newly acquired data sources. Constructed trail maps, trained on publicly available GPS data, are compared against a ground truth map from Open Street Map (OSM). Performance is evaluated based on completeness, accuracy, and topological correctness. Testing on sparse networks with minimal GPS noise suggests favorable performance.
- 27 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 37-42
AI Assisted Trail Map Generation based on Public GPS Data
Hiking trail maps are typically created manually by survey, a time-consuming process. This process is expensive and must be repeated over time to improve accuracy. This paper proposed an inexpensive, automatic, and accurate trail network generation method from anonymous public GPS data utilizing a growing self-organizing map (GSOM). This technique does not rely on sequential GPS traces to learn network topology, unlike other approaches. Tuning several hyper-parameters can adjust this process for datasets and networks with unique characteristics. Reconstruction and adaption are also possible based on newly acquired data sources. Constructed trail maps, trained on publicly available GPS data, are compared against a ground truth map from Open Street Map (OSM). Performance is evaluated based on completeness, accuracy, and topological correctness. Testing on sparse networks with minimal GPS noise suggests favorable performance.
- Open Access
- 16 March 2023
- Published by SPIE-Intl Soc Optical Eng
- p. 129
Miniaturization and geometric optimization of SteamVR active optical trackers
SPIE Digital Library Proceedings
- 16 March 2023
- Published by SPIE-Intl Soc Optical Eng
- p. 129
Open AccessMiniaturization and geometric optimization of SteamVR active optical trackers
SPIE Digital Library Proceedings
- 1 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 465-470
An Edge-based Real-Time Object Detection
This paper looks at performance bottlenecks of real-time object detection on edge devices. The "You only look once v4" (YOLOv4) is currently one of the leading state-of-the-art models for real-time object detection, and its tiny version: YOLOv4-tiny, is designed for edge devices. To improve object detection accuracy without sacrificing detection speed, we propose an object detection method based on YOLOv4-tiny and VGG-Net. First, we implement the mosaic data augmentation and Mish activation function to increase the generalization ability of the proposed model, making it more robust. Secondly, to enhance the richness of the features extracted, an extra 3x3 convolution layer is added in a way that two successive 3x3 convolutions are used to obtain 5x5 receptive fields. This would enable us to extract global features in the first CSP (Cross Stage Partial Network) Block and restructure the connections of the subsequent layers to have the same effect on the next CSP blocks. Evaluation results show that the proposed model has comparable performance and memory footprint but significantly greater accuracy than YOLOv4-tiny. Also, the proposed tiny model has similar performance to YOLOv4-tiny, and improves accuracy with much lower memory overhead, which makes it an ideal solution for real-time object detection, especially on edge devices.
- 1 December 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 465-470
An Edge-based Real-Time Object Detection
This paper looks at performance bottlenecks of real-time object detection on edge devices. The "You only look once v4" (YOLOv4) is currently one of the leading state-of-the-art models for real-time object detection, and its tiny version: YOLOv4-tiny, is designed for edge devices. To improve object detection accuracy without sacrificing detection speed, we propose an object detection method based on YOLOv4-tiny and VGG-Net. First, we implement the mosaic data augmentation and Mish activation function to increase the generalization ability of the proposed model, making it more robust. Secondly, to enhance the richness of the features extracted, an extra 3x3 convolution layer is added in a way that two successive 3x3 convolutions are used to obtain 5x5 receptive fields. This would enable us to extract global features in the first CSP (Cross Stage Partial Network) Block and restructure the connections of the subsequent layers to have the same effect on the next CSP blocks. Evaluation results show that the proposed model has comparable performance and memory footprint but significantly greater accuracy than YOLOv4-tiny. Also, the proposed tiny model has similar performance to YOLOv4-tiny, and improves accuracy with much lower memory overhead, which makes it an ideal solution for real-time object detection, especially on edge devices.
- 30 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Framework for Scalable Content Development in Hands-On Virtual and Mixed Reality Science Labs
Authentic hands-on laboratory research is essential for undergraduate STEM education. Yet the tactile authenticity required to impact affective, cognitive, or psychomotor learning outcomes associated with laboratory training remains underexplored. Virtual and mixed reality (VR/MR) have enabled increasingly realistic hands-on STEM training experiences. However, they still lack authenticity with regard to user manipulation of fully functional and realistic laboratory tools, analysis of realistic (i.e. user-acquired) noisy data, and the application of critical thinking skills to draw conclusions from such noisy (and possible faulty) data. Here we present efforts to develop such an approach while also providing faculty content experts tools for code-free customization of VR/MR training experiences via structured spreadsheets. This approach enables nuanced real-time user feedback on laboratory skills such as proper pipetting or sterile technique which are otherwise difficult to provide. It also offers complete safety from chemical, biological, and radiological hazards and is more cost-effective than a traditional lab. This Hands-On Virtual-Reality (HOVR) Lab platform is uniquely enabling and will be valuable in the physical and life sciences for both research and instructional applications.
- 30 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Framework for Scalable Content Development in Hands-On Virtual and Mixed Reality Science Labs
Authentic hands-on laboratory research is essential for undergraduate STEM education. Yet the tactile authenticity required to impact affective, cognitive, or psychomotor learning outcomes associated with laboratory training remains underexplored. Virtual and mixed reality (VR/MR) have enabled increasingly realistic hands-on STEM training experiences. However, they still lack authenticity with regard to user manipulation of fully functional and realistic laboratory tools, analysis of realistic (i.e. user-acquired) noisy data, and the application of critical thinking skills to draw conclusions from such noisy (and possible faulty) data. Here we present efforts to develop such an approach while also providing faculty content experts tools for code-free customization of VR/MR training experiences via structured spreadsheets. This approach enables nuanced real-time user feedback on laboratory skills such as proper pipetting or sterile technique which are otherwise difficult to provide. It also offers complete safety from chemical, biological, and radiological hazards and is more cost-effective than a traditional lab. This Hands-On Virtual-Reality (HOVR) Lab platform is uniquely enabling and will be valuable in the physical and life sciences for both research and instructional applications.
- 1 December 2021
- Published by IOP Publishing
Embedded intelligence for tracking facial expressions
- Open Access
- 25 November 2021
- Published by Institution of Engineering and Technology (IET) in IET Computers & Digital Techniques
- Vol. 16 (1) , 10-18
An embedded intelligence engine for driver drowsiness detection
Motor vehicle crashes involving drowsy driving are huge in number all over the world. Many studies revealed that 10%–30% of crashes are due to drowsy driving. Fatigue has costly effects on the safety, health, and quality of life. This drowsiness of drivers can be detected using various methods, for example, algorithms based on behavioural gestures, physiological signals and vitals. Also, few of them are vehicle based. Drowsiness of drivers was detected based on steering wheel movement and lane change patterns. A pattern is derived based on slow drifting and fast corrective steering movement. A prototype that detects the drowsiness of an automobile driver using artificial intelligence techniques, precisely using open‐source tools like TensorFlow Lite on a Raspberry Pi development board, is developed. The TensorFlow model is trained on images captured from the video with the help of object detection using cascade classifier. In order to have a better accuracy, an Inception v3 architecture is used in pre‐training the model with the image dataset. The final model is created and trained using long short‐term memory and then the final TensorFlow model is converted to TensorFlow Lite model and this Lite model is used on Raspberry Pi board to detect the drowsiness of drivers. The results are comparable with desktop‐based results in the literature.
- 25 November 2021
- Published by Institution of Engineering and Technology (IET) in IET Computers & Digital Techniques
- Vol. 16 (1) , 10-18
Open AccessAn embedded intelligence engine for driver drowsiness detection
Motor vehicle crashes involving drowsy driving are huge in number all over the world. Many studies revealed that 10%–30% of crashes are due to drowsy driving. Fatigue has costly effects on the safety, health, and quality of life. This drowsiness of drivers can be detected using various methods, for example, algorithms based on behavioural gestures, physiological signals and vitals. Also, few of them are vehicle based. Drowsiness of drivers was detected based on steering wheel movement and lane change patterns. A pattern is derived based on slow drifting and fast corrective steering movement. A prototype that detects the drowsiness of an automobile driver using artificial intelligence techniques, precisely using open‐source tools like TensorFlow Lite on a Raspberry Pi development board, is developed. The TensorFlow model is trained on images captured from the video with the help of object detection using cascade classifier. In order to have a better accuracy, an Inception v3 architecture is used in pre‐training the model with the image dataset. The final model is created and trained using long short‐term memory and then the final TensorFlow model is converted to TensorFlow Lite model and this Lite model is used on Raspberry Pi board to detect the drowsiness of drivers. The results are comparable with desktop‐based results in the literature.
- 1 June 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 71-76
Partially Homomorphic Encryption Scheme for Real-Time Image Stream
With over 20 billion Internet of Things (IoT) devices connected today, having the possibility to implement a low-cost security camera is now an option. Camera vendors can provide software to enable remote viewing anywhere around the world. However, the problem arises in the security measures taken. By only providing encrypted services to authenticate a user by login credentials, there still is a vulnerability in the raw stream. Newer types of intrusions like spoofing or man-in-the-middle attacks mean that once they have breached the account, they are able to gain full viewing access to the camera stream. However, by encrypting the stream of these IoT cameras, we have a chance to truly retain privacy and ownership of our data. With IoT devices being limited with on-board memory and CPU speeds, this can be a challenge. To mitigate and offset any CPU-intensive work, we will encrypt the data between our cloudlet device and our edge device by using a partially homomorphic encryption scheme called "Elgamal encryption". This CPU offloading allows the cloudlet to transmit streams at a much higher frame-rate. The proposed system will utilize a Raspberry Pi 4, for our cloudlet with our desktop being the edge.
- 1 June 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 71-76
Partially Homomorphic Encryption Scheme for Real-Time Image Stream
With over 20 billion Internet of Things (IoT) devices connected today, having the possibility to implement a low-cost security camera is now an option. Camera vendors can provide software to enable remote viewing anywhere around the world. However, the problem arises in the security measures taken. By only providing encrypted services to authenticate a user by login credentials, there still is a vulnerability in the raw stream. Newer types of intrusions like spoofing or man-in-the-middle attacks mean that once they have breached the account, they are able to gain full viewing access to the camera stream. However, by encrypting the stream of these IoT cameras, we have a chance to truly retain privacy and ownership of our data. With IoT devices being limited with on-board memory and CPU speeds, this can be a challenge. To mitigate and offset any CPU-intensive work, we will encrypt the data between our cloudlet device and our edge device by using a partially homomorphic encryption scheme called "Elgamal encryption". This CPU offloading allows the cloudlet to transmit streams at a much higher frame-rate. The proposed system will utilize a Raspberry Pi 4, for our cloudlet with our desktop being the edge.
- 10 September 2020
- Published by World Scientific Pub Co Pte Ltd in Journal of Circuits, Systems and Computers
- Vol. 30 (04)
A Distributed Smart Camera System Based on an Edge Orchestration Architecture
Distributed cameras have been used widely for real-time image recognition. There are two main approaches in distributed camera systems: (1) The cameras are equipped with a powerful high-end processor for local image processing, (2) Low-cost cameras with resource-constrained processors are used for capturing the images and transferring them to a cloud server for classification purposes. The first approach is costly and not scalable. The second approach is too slow for real-time object detection due to the transfer delays to a remote server. These problems exacerbate in multi-view image recognition, where a central platform is required for collective image processing of multiple images from the same scene. For this purpose, typically, a cloud server is used, which does not meet real-time recognition, network bandwidth, scalability and power consumption constraints of such systems. This paper proposes hierarchical neural network structures that can be realized in an edge orchestration architecture at different levels, i.e., cameras, edge devices, and cloud servers to enable deep learning capabilities for real-time multi-view image recognition. This would enable us to detect objects in the proximity of multiple cameras and transfer data for deeper layer processing to the cloud server and training of all connected edge devices and cameras.
- 10 September 2020
- Published by World Scientific Pub Co Pte Ltd in Journal of Circuits, Systems and Computers
- Vol. 30 (04)
A Distributed Smart Camera System Based on an Edge Orchestration Architecture
Distributed cameras have been used widely for real-time image recognition. There are two main approaches in distributed camera systems: (1) The cameras are equipped with a powerful high-end processor for local image processing, (2) Low-cost cameras with resource-constrained processors are used for capturing the images and transferring them to a cloud server for classification purposes. The first approach is costly and not scalable. The second approach is too slow for real-time object detection due to the transfer delays to a remote server. These problems exacerbate in multi-view image recognition, where a central platform is required for collective image processing of multiple images from the same scene. For this purpose, typically, a cloud server is used, which does not meet real-time recognition, network bandwidth, scalability and power consumption constraints of such systems. This paper proposes hierarchical neural network structures that can be realized in an edge orchestration architecture at different levels, i.e., cameras, edge devices, and cloud servers to enable deep learning capabilities for real-time multi-view image recognition. This would enable us to detect objects in the proximity of multiple cameras and transfer data for deeper layer processing to the cloud server and training of all connected edge devices and cameras.