About Me

Hadi Sadoghi Yazdi Resume

Areas of Expertise

🔶Pattern recognition	🔶Machine learning	🔶Machine vision
🔸 Data Analytics	🔸 Signal processing	🔸 Real-time data analysis
🔸 Image processing	🔸 Social network data analysis	🔸 Adaptive network
🔸 Optimization problems	🔸 Remote sensing	🔸 Wireless communication
🔸 Tracking	🔸 Speech processing	🔸 Interference cancellation
🔸 Deep learning	🔸 Microsoft AI developer	🔸 NLP (Natural language processing)
🔸 Bioinformatic (DNA methylation Analysis)	🔸 Big data

Technical Proficiencies

C, Matlab, Python, Google Colab, Jupyter Book, Jupyter Notebook, Git

Some Experiences

Classification of EEG Signal This project was carried out in Ferdowsi University of Mashhad, Iran. The product of this project is in the form of MATLAB coding based on artificial intelligence, which is able to use brain EEG signal to categorize different disease cases. In the form of the developer of the existing algorithms and the implementation of the proposed algorithm, I was engaged in this work in the form of a team of 3 people. Finally, by using existing classification methods and improving their performance, the output of this project reached an acceptable accuracy.

MRI Image segmentation This project was carried out in Ferdowsi University of Mashhad, Iran. The product of this project is in the form of Python coding based on artificial intelligence, which is able to segment the obtained MRI image of the human brain and categorize the types of brain tissue. In this project, this team was responsible for coding and implementing the algorithms, and the final result was the high accuracy of the algorithm compared to the existing ones.

Hyperspectral Image Segmentation This project was carried out in Ferdowsi University of Mashhad, Iran. The product of this project is in the form of coding based on artificial intelligence, which is able to prepare hyperspectral images obtained from sensing satellites for segmentation using a pre-processing block and then apply the segmentation method to the image. This pre-processing block based on the noise-resistant Kernel Density Estimation (KDE) method is able to reduce the effect of noisy labels and thus significantly increase the accuracy in the segmentation of different regions in the hyperspectral image.

Interactive wall As a senior researcher, Using a 3D camera such as Kinect to obtain 3D images online, plus a processor whose task is to perform image processing algorithms on the obtained images. For example, this system was developed in order to create entertainment for children, in which the child is interactively engaged in games designed on the wall. One of the existing challenges was determining and estimating the depth of objects, determining the effect of camera tilt on images, and calibrating the camera. The programming language for the artificial intelligence engine was Python, and a software for initial settings and game type determination was C#. Also, the game was written with UNITY. This system can be used for education, language and math education and creating activity in entertainment for children with autism.

Facebook and 500px social networks image labeling This project was carried out at Ferdowsi University of Mashhad, Iran, which has one of the best artificial intelligence groups among Iranian universities. The product of this project is in the form of coding based on artificial intelligence, which is able to label and categorize images based on images collected from Facebook and 500px social networks for applications such as tourist. In this project, the robust spectral clustering method was used to classify the features extracted from the AlexNet deep network, which significantly increased the accuracy of semantic labeling.

Traffic scene analysis As a designer of algorithms for plural behavior of driver modeling. In this project, the goal was to use artificial intelligence algorithms learning in predicting and modeling the driver’s behavior.

Vehicle Identifier Project Garmsar University, Garmsar, Iran, 2003-2004 As a designer of algorithms for identifying types of vehicles by fusion of two views of image of vehicle. As a designer of algorithms to identify types of vehicles by merging two views of the vehicle image. In this project, it was done using image processing algorithms, in which the type of car was identified by using two images taken by the camera from two different views of the car. Here, it was necessary to prepare a comprehensive library of images of all types of cars. The performance of the final product was acceptable on all types of cars.

Vision Based Tracker Project As a senior researcher, in study on different types of vision-based tracker systems. In this project, tracking based on video and processing its different frames was done in order to track a target in different frames. Artificial intelligence were used in this field.

Positioning of users in the cellular network, 2022 The purpose of this project was to locate users based on applied algorithms in this field, which was carried out at Ferdowsi University of Mashhad, Mashhad, Iran. This project works based on the power level received from the mobile station in closed environments. The purpose of this project is to help the handoff in the cellular network based on user categories. Based on this, users whose received power level is lower than a provincial limit need to be transferred to another base station for them. In this approach, spectral clustering algorithm was used to classify mobile stations.

Coach simulation As a Co-researcher, Coach simulation (movement comparison), for Smart Coach, Estonia. A team was working on this project and I did only the engineering intelligence part of it. It was written in Python for the core of artificial intelligence and in the form of API and with Flask. An Android application was also prepared for it. Coach performs a set of sports exercises and upload the video to the application, and also add the beginning and end of the movements and the type of exercises. Users choose their desired exercises, perform them and take a video of their exercise with their mobile phone and upload the video to the application. This video related to the user is analyzed and it is given a score based on the correctness of the movements performed. Points are given based on the comparison of this video with the trainer’s video. During the processing process, first the key points of body movements are extracted from the instructor’s video and then the corresponding points are compared with them from the user’s video frames. Finally, based on this comparison, the user’s movement is scored.

Camel GPS Tracker Application (Android), February 2022/ Camel GPS Tracker, February 2022/ SAREBANYAR App for camel husbandry and management, April 2020 As a Teammate, This project was performed for Animal Science Research Institute of IRAN (ASRI), Project Management, Camel GPS Tracker and Data Management and weight prediction Application (Android), Machine learning (Camel Weight Estimation), Link: sarebanyar.asri.ir/login. It was an application for camel data management that has both data management and data registration. The data is recorded in the management panel server, this data includes the number of animals in each farm, the number of males and females, age, etc. This information after aggregation was used for various applications. For example, weighing a camel is a challenge that can be done by processing the image with a camera or by measuring different parts of the body. In this project, it was finally done based on the dimensions of different body parts and by applying regression (SVR with polynomial kernel). By using this application, instead of weighing with a scale and applying stress to the camel, by measuring specific parts of the camel’s body and entering it into the application, farmers can get the approximate weight of the camel. Also, this application determines the location of the animal based on GPS data.

Student Sectioning Project, Sabzevar university, Iran 2006-2007 As a research assistant, this project was done using new fuzzy student sectioning approach.

Researcher and Developer (R&D) in Madiran (LG), Tehran, Iran
2004-2005 As researcher, Doing research in ESD( Electrostatic Discharge Control), Vision-based board fault finder, Designing LCD controller.

Presenting of OCR System in Tarbiat Modares University, Tehran, Iran
2004-2005 Developing of software for Optical Character Recognition using MATLAB software

OCR system, Hoda System Development Company, Iran 2004-2005 As a Director of Software Design for digit recognition for OCR system

Patient Recognizer Project Tehran University, Tehran, Iran 2003-2004 As a designer, developing algorithm and offering dentist patient recognizer based on new Bayesian rule-based system

ESD Projects, LG company (Madiran in Iran) 2000-2001 As a senior position in initializing and designing of ESD control

Teaching Experience

Education

🔸 PhD in Electronic Engineering, Tarbiat Modares University, Tehran	2005
🔸 Master of Electronic Engineering, Tarbiat Modares University, Tehran	1996
🔸 Bachelor of Electronic Engineering, Ferdowsi University of Mashhad, Tehran	1994

PRODUCTS and PATENTS

My group has many patents in the field of artificial intelligence and its application in industries, such as building a smart future, building a smart wall; Smart fire detection system, smart wall, facial recognition system for implementation in public places, etc. Some of these products are in the following:

🔶 Information management system on smart mirror	🔶 Heat maps: Density estimation
🔸 Interactive smart wall (3D camera)	🔸 Smart UAV system
🔸 Digital information panel	🔸 Smart emotion detection system
🔸 Smart camera on automobiles	🔸 Cancer diagnosis kit
🔸 Smoke and fire detection System	🔸 Smart parking
🔸 Traffic light control system (astronomical clock)	🔸 Smart monitoring disorder
🔸 Software development	🔸 Smart microscope
🔸 Laboratory hot plate stirrer device	🔸 Smart coach
🔸 Locator, transmitter and recorder kit of firemen’s vital signs	🔸 Pollution recording system by moving sensors
🔸 New method in the portfolio management of Stock Exchange, based on behavior of stocks	🔸 Smart system for detecting the abnormal behavior of bus drivers
🔸 Smart system for interacting Diabetic patients with their doctors	🔸 Designing A monitoring system for laboratory mice sports treadmill
🔸 Combining INS/GPS to determine location characteristics

PUBLICATIONS

He has more than 300 ISI articles published in prestigious journals of the world (Link).

Languages

| Farsi - Mother Tongue | |