Osman Tanko's Projects
An image segmentation Model that uses the U-Net Architecture to segment tumours from CT-Scan images of the breast.
data transmission over sound modulator-demodulator model
Byte is framework that faciliates data transmission via audible sound waves. This project is the web implemenation. It takes a messagefrom the user, encodes it into binary and then modulates it into sound
A website which uses webscraping to give you information about all countries in the world as well as their twitter trends
This project uses AI to fix blur images of faces. This uses a 5-layered UNet architecture which hallucinates facial features to sharpen the image. This be used recreationally or aid in forensic investigation to identify faces
This repository features a custom-built decoder-only language model (LLM) with a total of 37 million parameters š„. I train the model to be able to ask question from a given context
This is my implementation of a Multinomial Naive Bayes Classifier algorithm. The code predicts the class label of an input sequence based on its features. Therefore, given a dataset, the algorithm fits accurately and is able to make predictions based a sequence of features as input.
A python tool that checks to find exposed authentication keys in a project and throws a warning. This prevents the sharing of sensitive data and enforces the use of best paractices like saving authentication keys and passwords to .env files.
This is a notebook detailing the workflow followed in order to predict Loan Defaulters from a dataset of over 100,000 rows of customer information. The best model, Random Forest, had an accuracy of 100%
Web app that listens to sound and coverts that to to musical note
A demo chat app with which allows you to download videos on youtube
Simplified 2D NIH Pancreas dataset
This is an implementation of a simple Vision Transformer (ViT) for a dummy classification task of predicting whether a person is wearing a hat or not. š©š¤ š
A python Machine learning model designed for AI and Audio engineers to easily filter out vocal melodies from background instrumentals. This is effective for collecting and generating melodies of specific type, for example a specific musician. Other applications include speech denoising, and voice recognition