In this paper, we are presenting a new surface signature-based representation that is orientation... more In this paper, we are presenting a new surface signature-based representation that is orientation-independent and can be used to match and align surfaces under rigid transformation. The proposed scheme represents the surface patches in terms of their signatures. The surface signatures are formed as extended polar maps using the neighbours of each surface patch. Correlation of the maps is used to establish point correspondences between two views; from these correspondences a rigid transformation that aligns the views is calculated. The effectiveness of the proposed scheme is demonstrated through several registration experiments.
2021 17th International Computer Engineering Conference (ICENCO)
In this paper, we use the SlowFast Networks developed by the Facebook research team to enhance th... more In this paper, we use the SlowFast Networks developed by the Facebook research team to enhance the accuracy of dynamic sign language recognition. Firstly, we prepared the Word-Level American Sign Language (WLASL) dataset so each sign can be considered an action. We used the pre-trained SLOWFAST_8×8_R50 model provided on the official PySlowFast Github repository to initialize the weights of our model and fine-tune using the WLASL dataset and performed a param-eter sweeping to fit the Dynamic Sign Language Recognition task. Through this transfer learning approach, we managed to introduce a new state-of-the-art accuracy on the WLASL300 (300 words e.g., 300 classes) dataset with an improvement of 23.2 % top-1 accuracy compared to the previous state-of-the-art introduced in the WLASL paper using an I3D model. The top-1 accuracy was improved from 56.14% to 79.34% and the top-5 accuracy from 79.94% to 90.31 %.
Indonesian Journal of Electrical Engineering and Computer Science, 2022
In computer vision, one of the most difficult problems is human gestures in videos recognition Be... more In computer vision, one of the most difficult problems is human gestures in videos recognition Because of certain irrelevant environmental variables. This issue has been solved by using single deep networks to learn spatiotemporal characteristics from video data, and this approach is still insufficient to handle both problems at the same time. As a result, the researchers fused various models to allow for the effective collection of important shape information as well as precise spatiotemporal variation of gestures. In this study, we collected the dynamic dataset for twenty meaningful words of Arabic sign language (ArSL) using a Microsoft Kinect v2 camera. The recorded data included 7350 red, green, and blue (RGB) videos and 7350 depth videos. We proposed four deep neural networks models using 2D and 3D convolutional neural network (CNN) to cover all feature extraction methods and then passing these features to the recurrent neural network (RNN) for sequence classification. Long sho...
Due to the enormous use of mobile applications and the wide spread of location-based services, as... more Due to the enormous use of mobile applications and the wide spread of location-based services, as Foursquare, google maps, Facebook check-ins, it became a must to focus on studying these data and its impact on our social norms. In this paper, we are tackling the location novelty problem, which evaluates the user’s curiosity to explore new places. In order to maintain a better service and offer new services, such as recommending new places, optimizing marketing campaigns, we conducted these experiments to classify the next check-ins to be either Novel or regular. We can predict the novelty of the next Point of Interest (POI) up to 82%, by extracting different types of features, in space and time, and using boosting trees.
2020 5th IEEE International Conference on Recent Advances and Innovations in Engineering (ICRAIE), 2020
The electric Load is affected by various factors such as economic, social, and meteorological fac... more The electric Load is affected by various factors such as economic, social, and meteorological factors. This classification simplifies the studying of the correlation between these factors. It also provides a useful reference for researchers to pick up the best elements for their case according to the forecasting period; short, medium, or long term forecasting. This work introduces a comprehensive study of the factors that affect load forecasting in short, medium, and long-term load forecasting. Correlational model is applied to assess the relationship among parameters in different time horizons. The result provides two critical things to notes. Firstly, there are direct and indirect effects for some parameters based on the timeframe, and secondly, there is a significant accumulative effect of some parameters.
A novel approach is proposed to obtain a record of the patient's occlusion using computer vis... more A novel approach is proposed to obtain a record of the patient's occlusion using computer vision. Data acquisition is obtained using intra-oral video camera. The technique utilizes shape from shading to extract 3D information from 2D views of the jaw, and a novel technique for 3D data registration using genetic algorithms. The resulting 3D model can be used for diagnosis, treatment planning, and implant purposes. The overall purpose of this research is to develop a model-based vision system for orthodontics to replace traditional approaches. This system will be exible, accurate, and will reduce the cost of orthodontic treatments.
Handbook of Research on Deep Learning-Based Image Analysis Under Constrained and Unconstrained Environments, 2021
Object classification problem is essential in many applications nowadays. Human can easily classi... more Object classification problem is essential in many applications nowadays. Human can easily classify objects in unconstrained environments easily. Classical classification techniques were far away from human performance. Thus, researchers try to mimic the human visual system till they reached the deep neural networks. This chapter gives a review and analysis in the field of the deep convolutional neural network usage in object classification under constrained and unconstrained environment. The chapter gives a brief review on the classical techniques of object classification and the development of bio-inspired computational models from neuroscience till the creation of deep neural networks. A review is given on the constrained environment issues: the hardware computing resources and memory, the object appearance and background, and the training and processing time. Datasets that are used to test the performance are analyzed according to the images environmental conditions, besides the...
2016 Eighteenth International Middle East Power Systems Conference (MEPCON), 2016
The electric load influenced by different factors such as meteorological, economics, and casual f... more The electric load influenced by different factors such as meteorological, economics, and casual factors according to demographic location, time, and the human behavior. This classification simplified studying the correlation between those factors. It is important to select the right set of parameters that affect the forecasting. Selecting irrelevant parameters will require additional computation time and may not improve the forecasting accuracy. This work introduces the effect of electrical load factors in short term load forecasting. In this work, several factors (temperature, due temperature, wind, and humidity) are applied to ANN to understand its impact on electric load forecasting of Northern Cairo. From the experimental results, we show that MAPE, RMSE, and MAE are decreased by more than half after using the proposed model.
2017 IEEE 7th Annual Computing and Communication Workshop and Conference (CCWC), 2017
Since the early 20th century, Electricity is a necessity and essential to modern life. Producing ... more Since the early 20th century, Electricity is a necessity and essential to modern life. Producing and delivering electricity to 7.4 billion people is one of the most complex challenges in this decade. It is essential to forecast the energy load at different intervals. Several methods have been developed to carry out those predictions. Examples of such methods are ARIMA, regression and artificial Neural Network (ANN). This work investigates the usage of adaptive Fourier series in enhancing the prediction accuracy of ARIMA and ANN models. Those models along with their enhanced ones are applied to forecasting hourly Peak load data of The Electric Reliability Council of Texas (ERCOT). The experimental results demonstrate that using the adaptive Fourier series enhances the prediction power of the ARIMA and ANN models. The RMSE, MAE and MAPE are enhanced by increasing the number of harmonics required for a specific seasonal component.
2016 Eighteenth International Middle East Power Systems Conference (MEPCON), 2016
The Electric load supply industry requires forecasts with lead times that range from short terms ... more The Electric load supply industry requires forecasts with lead times that range from short terms (a few hours, or days ahead) to long terms (up to 10 years ahead). Load forecasting is a complex task because of high non-linearity relation among load variables and load exhibits several levels of seasonality. This paper presents the effect of Hijri calendar on load forecasting. Hijri calendar could be useful on different time horizon. In Middle East region, most countries are influenced by different Islamic Hijri calendar (i.e. Ramadan, religious holidays, weekend, Friday, and different workday). Only during Ramadan, many casual events occurred such as workday changing, fasting, and more usage for air conditioning during the whole month. The peak load in Ramadan differs from other months on the workdays. Based on our experiments when forecasting north Cairo electric load, the forecasting accuracy is increased when including casual Hijri events. Moreover, including casual Hijri events results in a large decrease in the forecasting errors; RMSE, MAE and MAPE.
2016 12th International Computer Engineering Conference (ICENCO), 2016
Despite License Plate Recognition is mainly regarded as a solved problem; most of the techniques ... more Despite License Plate Recognition is mainly regarded as a solved problem; most of the techniques have been mainly developed for specific country or special formats which can strictly limits their applicability. There have been extensive studies of license plate detection since the 70s. The suggested approaches have difficulties in processing high-resolution imagery in real-time. This paper presents a novel algorithm for real-time automatic multi-style license plate detection in videos. The proposed algorithm can detect in a real time multiple license plates with various sizes in unfamiliar and complex environment. In this system, candidate plate regions are extracted using a preprocessing function to increase accuracy while decreasing computational time. Then a tree of LBP-based cascade classifiers is used to classify the candidate plate regions into one of the learned style. The proposed approach has been applied to Egyptian license plates with four different plate styles. The proposed approach achieved a success rate of 94% at 25 frames/sec using a moderate laptop.
In this paper, we are presenting a new surface signature-based representation that is orientation... more In this paper, we are presenting a new surface signature-based representation that is orientation-independent and can be used to match and align surfaces under rigid transformation. The proposed scheme represents the surface patches in terms of their signatures. The surface signatures are formed as extended polar maps using the neighbours of each surface patch. Correlation of the maps is used to establish point correspondences between two views; from these correspondences a rigid transformation that aligns the views is calculated. The effectiveness of the proposed scheme is demonstrated through several registration experiments.
2021 17th International Computer Engineering Conference (ICENCO)
In this paper, we use the SlowFast Networks developed by the Facebook research team to enhance th... more In this paper, we use the SlowFast Networks developed by the Facebook research team to enhance the accuracy of dynamic sign language recognition. Firstly, we prepared the Word-Level American Sign Language (WLASL) dataset so each sign can be considered an action. We used the pre-trained SLOWFAST_8×8_R50 model provided on the official PySlowFast Github repository to initialize the weights of our model and fine-tune using the WLASL dataset and performed a param-eter sweeping to fit the Dynamic Sign Language Recognition task. Through this transfer learning approach, we managed to introduce a new state-of-the-art accuracy on the WLASL300 (300 words e.g., 300 classes) dataset with an improvement of 23.2 % top-1 accuracy compared to the previous state-of-the-art introduced in the WLASL paper using an I3D model. The top-1 accuracy was improved from 56.14% to 79.34% and the top-5 accuracy from 79.94% to 90.31 %.
Indonesian Journal of Electrical Engineering and Computer Science, 2022
In computer vision, one of the most difficult problems is human gestures in videos recognition Be... more In computer vision, one of the most difficult problems is human gestures in videos recognition Because of certain irrelevant environmental variables. This issue has been solved by using single deep networks to learn spatiotemporal characteristics from video data, and this approach is still insufficient to handle both problems at the same time. As a result, the researchers fused various models to allow for the effective collection of important shape information as well as precise spatiotemporal variation of gestures. In this study, we collected the dynamic dataset for twenty meaningful words of Arabic sign language (ArSL) using a Microsoft Kinect v2 camera. The recorded data included 7350 red, green, and blue (RGB) videos and 7350 depth videos. We proposed four deep neural networks models using 2D and 3D convolutional neural network (CNN) to cover all feature extraction methods and then passing these features to the recurrent neural network (RNN) for sequence classification. Long sho...
Due to the enormous use of mobile applications and the wide spread of location-based services, as... more Due to the enormous use of mobile applications and the wide spread of location-based services, as Foursquare, google maps, Facebook check-ins, it became a must to focus on studying these data and its impact on our social norms. In this paper, we are tackling the location novelty problem, which evaluates the user’s curiosity to explore new places. In order to maintain a better service and offer new services, such as recommending new places, optimizing marketing campaigns, we conducted these experiments to classify the next check-ins to be either Novel or regular. We can predict the novelty of the next Point of Interest (POI) up to 82%, by extracting different types of features, in space and time, and using boosting trees.
2020 5th IEEE International Conference on Recent Advances and Innovations in Engineering (ICRAIE), 2020
The electric Load is affected by various factors such as economic, social, and meteorological fac... more The electric Load is affected by various factors such as economic, social, and meteorological factors. This classification simplifies the studying of the correlation between these factors. It also provides a useful reference for researchers to pick up the best elements for their case according to the forecasting period; short, medium, or long term forecasting. This work introduces a comprehensive study of the factors that affect load forecasting in short, medium, and long-term load forecasting. Correlational model is applied to assess the relationship among parameters in different time horizons. The result provides two critical things to notes. Firstly, there are direct and indirect effects for some parameters based on the timeframe, and secondly, there is a significant accumulative effect of some parameters.
A novel approach is proposed to obtain a record of the patient's occlusion using computer vis... more A novel approach is proposed to obtain a record of the patient's occlusion using computer vision. Data acquisition is obtained using intra-oral video camera. The technique utilizes shape from shading to extract 3D information from 2D views of the jaw, and a novel technique for 3D data registration using genetic algorithms. The resulting 3D model can be used for diagnosis, treatment planning, and implant purposes. The overall purpose of this research is to develop a model-based vision system for orthodontics to replace traditional approaches. This system will be exible, accurate, and will reduce the cost of orthodontic treatments.
Handbook of Research on Deep Learning-Based Image Analysis Under Constrained and Unconstrained Environments, 2021
Object classification problem is essential in many applications nowadays. Human can easily classi... more Object classification problem is essential in many applications nowadays. Human can easily classify objects in unconstrained environments easily. Classical classification techniques were far away from human performance. Thus, researchers try to mimic the human visual system till they reached the deep neural networks. This chapter gives a review and analysis in the field of the deep convolutional neural network usage in object classification under constrained and unconstrained environment. The chapter gives a brief review on the classical techniques of object classification and the development of bio-inspired computational models from neuroscience till the creation of deep neural networks. A review is given on the constrained environment issues: the hardware computing resources and memory, the object appearance and background, and the training and processing time. Datasets that are used to test the performance are analyzed according to the images environmental conditions, besides the...
2016 Eighteenth International Middle East Power Systems Conference (MEPCON), 2016
The electric load influenced by different factors such as meteorological, economics, and casual f... more The electric load influenced by different factors such as meteorological, economics, and casual factors according to demographic location, time, and the human behavior. This classification simplified studying the correlation between those factors. It is important to select the right set of parameters that affect the forecasting. Selecting irrelevant parameters will require additional computation time and may not improve the forecasting accuracy. This work introduces the effect of electrical load factors in short term load forecasting. In this work, several factors (temperature, due temperature, wind, and humidity) are applied to ANN to understand its impact on electric load forecasting of Northern Cairo. From the experimental results, we show that MAPE, RMSE, and MAE are decreased by more than half after using the proposed model.
2017 IEEE 7th Annual Computing and Communication Workshop and Conference (CCWC), 2017
Since the early 20th century, Electricity is a necessity and essential to modern life. Producing ... more Since the early 20th century, Electricity is a necessity and essential to modern life. Producing and delivering electricity to 7.4 billion people is one of the most complex challenges in this decade. It is essential to forecast the energy load at different intervals. Several methods have been developed to carry out those predictions. Examples of such methods are ARIMA, regression and artificial Neural Network (ANN). This work investigates the usage of adaptive Fourier series in enhancing the prediction accuracy of ARIMA and ANN models. Those models along with their enhanced ones are applied to forecasting hourly Peak load data of The Electric Reliability Council of Texas (ERCOT). The experimental results demonstrate that using the adaptive Fourier series enhances the prediction power of the ARIMA and ANN models. The RMSE, MAE and MAPE are enhanced by increasing the number of harmonics required for a specific seasonal component.
2016 Eighteenth International Middle East Power Systems Conference (MEPCON), 2016
The Electric load supply industry requires forecasts with lead times that range from short terms ... more The Electric load supply industry requires forecasts with lead times that range from short terms (a few hours, or days ahead) to long terms (up to 10 years ahead). Load forecasting is a complex task because of high non-linearity relation among load variables and load exhibits several levels of seasonality. This paper presents the effect of Hijri calendar on load forecasting. Hijri calendar could be useful on different time horizon. In Middle East region, most countries are influenced by different Islamic Hijri calendar (i.e. Ramadan, religious holidays, weekend, Friday, and different workday). Only during Ramadan, many casual events occurred such as workday changing, fasting, and more usage for air conditioning during the whole month. The peak load in Ramadan differs from other months on the workdays. Based on our experiments when forecasting north Cairo electric load, the forecasting accuracy is increased when including casual Hijri events. Moreover, including casual Hijri events results in a large decrease in the forecasting errors; RMSE, MAE and MAPE.
2016 12th International Computer Engineering Conference (ICENCO), 2016
Despite License Plate Recognition is mainly regarded as a solved problem; most of the techniques ... more Despite License Plate Recognition is mainly regarded as a solved problem; most of the techniques have been mainly developed for specific country or special formats which can strictly limits their applicability. There have been extensive studies of license plate detection since the 70s. The suggested approaches have difficulties in processing high-resolution imagery in real-time. This paper presents a novel algorithm for real-time automatic multi-style license plate detection in videos. The proposed algorithm can detect in a real time multiple license plates with various sizes in unfamiliar and complex environment. In this system, candidate plate regions are extracted using a preprocessing function to increase accuracy while decreasing computational time. Then a tree of LBP-based cascade classifiers is used to classify the candidate plate regions into one of the learned style. The proposed approach has been applied to Egyptian license plates with four different plate styles. The proposed approach achieved a success rate of 94% at 25 frames/sec using a moderate laptop.
Uploads