Short bio : I have more than 4 years of working experience in computer vision. Currently, I am postdoc-researcher at École nationale supérieure des mines de Saint-Étienne. Check out my publications, teaching, and projects for more information.
Oussama.location
Lyon, France
Oussama.education
Ph.D in Image processing using Artificial Intelligence (AI)
Master's in Automation and system's control
Oussama.contact
Oussama.resume
Oussama.skills
[C/C++, VB, Python, Matlab, Pytorch/Tensorflow, Tia Portal, Linux/windows os]
Oussama.interests
[AI, 3D Computer vision, Automation, Hacker spaces]
Oussama.languages
[english, french, arabic]
"Today, all you have to do is go online." — Bill Gates
Introducing an innovative online app that assesses the quality of 2D images using cutting-edge deep learning algorithms; a deep Vison Transformer (ViT) model.
All of the computing is done right on the edge of the client, which means you can easily use the app offline once it is loaded and evaluate images in a variety of formats, including JPEG, PNG, and others.
With this free app, you can be confident that your images will be evaluated precisely and accurately, allowing you to make informed decisions about the quality of your digital content. Whether you're a professional photographer, graphic designer, or just someone who wants to make sure their images are of high quality.
So why wait? Try our app today and experience the power of deep learning for yourself!
Left: the input partially distorted image. Right: the output quality map:
The first online app for assessing the quality of stereoscopic 3D images. The metric is based on a deep Convolutional Neural Network (CNN) and does not require pre-processing; it should work with any image resolution greater than 32 x 32 pixels, and it supports all image formats (e.g., BMP, PNG, JPG, GIF ...etc).
Click on the link below to test the app, please note that the app does not record or save any user's data/images but a feedback is appreciated.
The simplest way to deploy image processing applications online is to use javascript (on-edge execution). If there is an API and we only need to call the functions, it will be even easier and faster. I used the face-analysis API in this case to deploy online face analysis app from the user camera video stream. Please try the live demonstration (link below), the app does not record or save any user's data/images.
"The way to succeed is to double your failure rate." - Anonymous
Deep learning in industrial applications is becoming incredibly beneficial due to its promising performance. However, due to the high temperature of the metal coil, most metal coil production lines do not allow for fault detection sensors. However, we can detect and classify a deformation during the rolling of a hot metal coil using only a video stream. For that, a deep pre-trained CNN model has been used. Find the code in link below, written in Python using Tensorflow toolbox..
I have developed a software (PLC + HMI as seen in figures below), for Cemente Block machines. The production process has been maximized using multi-task processing. As a result, the production quantity is maximized and run-time matches the state-of-the-art records. Furthermore, The software provides real time animations and contains up to 30 alarms for critical situations. More than 100 receipts can be saved and executed by the software.
So far, the software is running on one machine for more than two years without any bugs.
I have developed a software (PLC + HMI) for concrete batching plants. The manufacturing process has been optimized using multi-task processing. As a result, we have recorded a state-of-the-art production time. The software includes up to 50 alarms for critical situations and provides real-time animations. In addition, more than 100 receipts can be saved and executed by the software.
It is running on two machines for more than three years without any bugs.
I have established an automatic pumping system with real time level indicators with more than 15 alarms and real time animations. (Automatic & Manual control via HMI panel)
The system is stable and runs perfectly on three stations for more than three years.
"Teaching is the highest form of understanding." - Aristotle
Teaching and coaching future engineers in the following courses:
❶ Information and communication technology (Word, Excel, Powerpoint ..etc)
❷ Algorithmic and programming using Blockly.
❸ Algorithmic and complexity using python language.
❹ Visual Basic for Application (VBA) programming language.
Teaching and coaching future engineers in the following courses:
❶ Algorithmic and coding using C language.
❷ Visual Basic for Application (VBA) programming language.
“The true sign of intelligence is not knowledge but imagination” - Albert Einstein
[1] O. Messai, A. Chetouani “End-to-End Deep Multi-Score Model for No-Reference Stereoscopic Image Quality Assessment”, in 2022 IEEE International Conference on Image Processing (ICIP)
[2] O. Messai, A. Chetouani, F. Hachouf, and Z. Ahmed Seghir, “3D Saliency guided Deep Quality predictor for No-Reference Stereoscopic Images”, in Neurocomputing Journal, January 06, 2022, Elsevier. [3] O. Messai, A. Chetouani, F. Hachouf, and Z. Ahmed Seghir, “Deep Quality evaluator guided by 3D Saliency for Stereoscopic Images”, in 26th International Conference on Human Vision and Electronic Imaging (HVEI), California, USA, January 18-21, 2021, p. 110, Society for Imaging Science and Technology (IST). [4] O. Messai, A. Chetouani, F. Hachouf, and Z. Ahmed Seghir, “No-reference Stereoscopic Image Quality Predictor using Deep Features from Cyclopean Image”, in 18th International Conference on Image Quality and System Performance (IQSP), California, USA, January 18-21, 2021, p. 297, Society for Imaging Science and Technology (IST). [5] O. Messai, F. Hachouf, and Z. Ahmed Seghir, “Adaboost Neural Network and Cyclopean View for No-reference Stereoscopic Image Quality Assessment,” Signal Processing: Image Communication, vol. 82, pp. 115772, March, 2020. [6] O. Messai, F. Hachouf, and Z. A. Seghir, “Automatic Distortion Type Recognition for Stereoscopic Images”, in International Conference on Advanced Electrical Engineering (ICAEE), Algiers, Algeria, November 19–21, 2019, pp. 1–4, IEEE.