Anonymization of legal cases (Fr) based on Flair embeddings

[Python] Additional works on Edward Choi's medGAN (generative adversarial network for electronic health records). In particular: boosting the prediction score using dataset augmentation.

Lightweight image converter and dataset augmentor

BigRocq is a utility, that takes a Rocq (former Coq) project as input and uses domain knowladge to increase a number of theorems in the dataset by a significant factor.

First-Prize winner in the Core Track of the 2022 OpenCV AI Competition

University MS Thesis Project, Controlling an avatar in a Virtual Environment via EEG Motor Imagery

A python library to perform patch augmentation.

Custom pedestrian detection based on yolov3.

learnig synthetic dataset generation (data augmentation) with tensorflow lib for computer vision

This model YOLOv5 🚀 is used to benchmark patch augmentation performance of patchmentation.

This datasets are used to benchmark patch augmentation performance of patchmentation.

This repository contains PyTorch implementation of various architectures and some helper functions for data augmentation, custom dataset building etc...

Synthetic Data Generation: Tabular & Medical Imaging. A comprehensive project focused on generating synthetic data for tabular datasets and medical imaging.

Developing Convolutional Neural Networks (CNN) for predicting whether a clothing item is classified as Casual, Formal or Smart Casual wear. Comparison analysis of a simple CNN and an improved neural network.

An LLM-powered dataset-aligned caption generator for downstream synthetic image generation.

simple tag expander/merger based on what's found within a tagged dataset

Exploration into augmenting datasets using GenAI by changing overrepresented classes into underrepresented classes.

This tool is designed to help balance class distributions in datasets, which is particularly useful for enhancing the performance of machine learning models affected by class imbalance.

Real-time Finnish traffic sign detection and classification using synthetic data, transfer learning, and advanced deep learning techniques.

Generate robot video variants with Cosmos and build augmented datasets.