Aymen BOUGUERRA's repositories
Unet-denoising-tool-for-textbook-images-UDM-
This tool will use a pre-trained Unet model to remove grids, grey background and noise from writing and equations in textbook type papers (with diffent types of grids)
Data-science-project-for-parcel-expedition-delays
An academic projet that I took a bit too seriously, the goal was to strat from a seemingly sterile dataset and forge it into a usable dataset with very strong features that would allow a machine/deep learning algorithme to learn and generelize the amount of time (in days) needed for the parcel expidition
Data_Science_Project_Expedition_Delay_Prediction
A project given by a very experienced data-scientist, the goal was to work on a dataset with seemingly useless features and extract useful ones from them. the focus was not on the models as they were simple regression models, the main work was about how to extract useful features that would have a big impact on increasing the model's accuracy.
Deep_Learning_For_Loan_Approval
This was a project that I worked on during my 4th year, the goal was to look for the best dataset that could be used for credit loan approval, we had to clean, feature engineer and then clean the dataset again as it had too much useless features (the dataset was not destined for this application) then applied a binary classification model.
Predict-online-shopping-intentions-using-machine-learning-and-deep-learning-approaches
This was a project that I worked on during my 4th year in an engineering school, the goal was to use machine and deep learning methods to not only predict if the operation will result in a purchase or not but also compare the different algorithms between them in terms of speed, accuracy and other metrics of performance..
Tool-for-making-noisy-images-and-their-masks-for-Unet-appliation-
While working on a Unet project, I created a program that can be used to add noise, a random grid (textbook) and a random shade of grey , this tool will output (depending on witch variation) combinations of two images the noisy image ut self and the clear one for the first variation (this one gave better results with Unet application) while the second variation will output the noisy image and the noise as its mask
UDM-TrOCRM
Inclusive mathematics education based on digital learning Project Reference : 2021-1-FR01-KA220-HED-000023486. This work is part of the IMEDiL (Inclusive Mathematics Education based on Digital Learning) program that seeks to create digital tools to help and assist the impaired persuing STEM education.