University Projects

A computer vision project that detect and locate objects within complex scenes. 

It begins by converting images of a singular object against a plain background and a cluttered scene containing multiple objects into black and white. 

Next, distinctive points of interest are identified within both images.

Feature extraction and comparison techniques are then employed to analyse similarities between these points in both images, enabling the precise localization of the object within the cluttered scene. 

Finally, a geometric transformation estimation process outlines the object's position amidst the surrounding elements.

A Mamdani type fuzzy logic system with six inputs and one output. 

It's designed to assist in investment decision-making based on various financial indicators. 

The inputs include metrics such as revenue growth, gross profit margin, and P/E ratio, each divided into linguistic terms like "negative," "average," and "high." 

The output, "Type of Investment," provides linguistic categories such as "avoid," "risky," "worthy," and "unicorn" based on the inputs' fuzzy logic analysis. 

The system's rules govern the mapping between input combinations and output categories, enabling nuanced investment recommendations.

I invested €100 in it's decision for testing purposes and ended up with 30% profit after one month (not an investing advice).

Made with Python using Tensorflow, NLTK and Numpy.

Its purpose is to have a free conversation with the user.

It operates by:
- accepting a sentence from the user,
- processes the natural language within it,
- and employs neural networks to determine the most suitable response.

Techniques such as tokenization, stemming, and bag-of-words are utilized for this purpose.

The user can:

draw with pencil, change pencil color, use eraser, change pencil density, draw shapes, change shapes' border and fill color,  select, move, rotate and resize shapes, undo change, select/deselect all shapes, delete selected shapes, fill background, make program semi-transparent, standard file menu with new, open, save as, exit. Files can be saved as png, jpg and bmp.

This project includes:

eBook list with search fields,
BGTOLL,
Weather forecast,
Currency converter,
Music / Movie / Superhero information APIs.

Drivers: View, add, update, and remove driver records.

Trucks: Track company trucks, including manufacturer, model, registration year, and vehicle specifications.

Partner Companies: Manage partner logistics companies and manufacturers associated with the cargo and drivers.

Advanced Search: Query the database with multi-field search logic to filter drivers, companies, and trucks dynamically.

Embedded Database: Runs out-of-the-box using an embedded H2 database file