IFGuard: Real-Time Anomaly Transaction Detection

The project, "IFGuard: Real Time Fraud Transaction Detection," is a sophisticated system designed to detect and score potential anomalous transactions in real-time. It leverages the power of the Isolation Forest algorithm for anomaly detection, and incorporates data streaming and distributed computi

StellarNet: CNN Architecture for Stellar Classification based on Spectral Characteristics

The StellarNet architecture is builded and fine tuned using the SDSS17 dataset in which due to the imbalance in the dataset, two data balancing techniques were applied: minimum class balance and Synthetic Minority Over-sampling Technique (SMOTE) balance. Both approaches yielded impressive results.

Real-Time Facial Emotion Recognition for Mental Health Therapy Sessions

This project utilizes MediaPipe, DeepFace, OpenCV, and Streamlit among other tools to detect emotions in real-time. This allows psychologists and psychiatrists to gain insights into their patients’ reactions to specific topics or stimuli, ensuring no detail is overlooked.

XChestNet: Model for Computer-Aided Chest XRay Diagnosis

XChestNet involves the creation of a CNN architecture using transfer learning and fine tunning on the Xception Engine to diagnose chest X-rays of healthy patients, those with pneumonia, and those with COVID-19, including also a streamlit application for diagnosis and image analysis

Intelligent Integrated System for Future Crop Recommendation: Advancing Sustainable Agriculture in Cuba.

The IIS is designed to recommend future crops in Cuba using climate forecasting and crop recommendation models. The aim is to enhance efficiency and productivity within the agricultural sector.

Fruits & Vegetables Freshness Classification

A solution for the multi-class classification of fresh and rotten fruits and vegetables is developed. Two distinct approaches are explored, Transfer Learning and CNN Architecture from scratch having both differents stronger points. Allowing to enhance food safety and reduce waste.

Google Finance Forecasting & Anomaly Detection

The project employs simple neural networks for forecasting and a suite of detectors for identifying various types of anomalies. These detectors include a Volatility, Persistent, Inter Quartile, Quantile, and Threshold.

Fashion MNIST Deep Convolutional Generative Adversarial Neural Network

A DCGAN to generate synthetic instances from the Fashion MNIST dataset, allowing to produce realistic images of fashion items.

Fast Facial Emotion Monitoring (FFEM) PyPi

Easy and Fast Facial Emotion Monitoring. This package is designed with user-friendliness in mind, making the complex task of FER accessible and straightforward for users.

Schneider Electric European Hackathon 2023: EcoForecast Revolutionizing Green Energy Surplus Prediction in Europe

Our solution for the Schneider Electric European Hackathon 2023 is a set of EcoForecast models using Prophet, to predict the surplus of green energy for various european countries.

NASA's Pale Blue Dot Mission: World Temperature Viewer

Our solution underscores the importance of data-driven decision-making in addressing the pressing issue of climate change by providing a platform that translates rich climate data into actionable insights, facilitating informed discussions and decisions about sustainable practices and policies.