Veer Singh's repositories
unideb_bsc_comp_sci_eng_state_exam
π£οΈ State Exam answers and guide for BSc. Computer Science Engineering, University of Debrecen. Use the website in the description.
OCR-with-Tesseract
βοΈ Contains 4 python modules. Basic OCR using Google's Tesseract on single image and pdf. Auto orientation correction for scanned docs. Auto noise type detection and reduction. Watermark and stain removal on scanned docs. These modules act as preprocessing tools for the best OCR results.
Recommender-System-using-Personal-Spotify-Data
πΆ 8 Machine Learning classification algorithms were used on my personal spotify data to create the ultimate playlist. Features were scaled for some algorithms, hyperparameters were tuned to achieve best results. Data was collected using Spotify Web API and then cleaned and labelled.
Chat-application
π¬ A WPF chat application with MySQL database. Lets users connect using personal accounts, send personal or public messages and also share files.
Custom-Object-Detection-with-TensorFlow-2
π΅οΈββοΈ A custom model was created using TensorFlow 2 on a novel dataset. Dataset consisted of 2,400 images and had an accuracy of 85%.
Custom-Object-Detection-with-TensorFlow-2-Lite-on-Raspberry-Pi
π A custom model was created using TensorFlow 2 Lite on a novel dataset. Dataset consisted of 2,400 images and had an accuracy of 85%. The final tests were done on a Raspberry Pi 4.
Linux-File-Security
π My notes for the IT Security course
Raspberry-Pi-Mini-Projects
π Mini projects performed on a Raspberry Pi 4 with Python.
Binary-Classification-CNN-with-Keras
π±πΆ Created a binary classifier CNN with Keras. Trained the model on AMD GPU using PlaidML as backend.
Predicting-Stroke-using-Various-Classification-Methods
π Code accompanying my bachelor's thesis. Find the download link to the thesis in the README.
signature-detection-and-extraction
ποΈ 3 open source signature detection and extraction modules were analyzed. 3 more modules were created by modifying and combining the open source modules, creating a module which outperformed all existing modules. A custom CNN was built using Keras to detect signatures and pre existing VGG16 CNN was modified to create a signature extraction model. Trained and tested on the Tobacco800 dataset.
Administrative-System-based-on-Neptun
π Meptun - Educational administration system for students. Desktop application written in Java using H2 database. UI was created using JavaFX. Project was completed in a team of 4 using Trello as project management software.
Business-Aggregator-Web-App
https://veer-singh.com/CSCI-571/zezima.html
Image-Analyser
π This android app lets the user pick an image from their gallery and then tries to identify some basic elements found in the image.
signature_detection
A package for signature detection
Veer-s-Expense-Report
πΈ This android app lets the user track my expenses incurred over a typical month.