My Work

Projects

Selected machine learning and AI projects

TFG: ArtAuthAI

Bachelor's Thesis - AI-powered artwork authentication system using deep learning to detect art forgeries and verify authenticity.

Key Highlights

  • Developed deep learning models for art style analysis
  • Achieved 94% accuracy in forgery detection
  • Implemented transfer learning with ResNet architecture
  • Created comprehensive dataset of artistic styles

Technologies

PythonTensorFlowComputer VisionCNNResNet

Outcome

Computer vision system trained and evaluated on a custom artwork dataset.

Deepfake Detection Pipeline

Master's Thesis - Advanced deepfake detection pipeline for both audio and video content, using custom-trained models by the research team.

Key Highlights

  • Achieved 98% accuracy in audio deepfake detection
  • Achieved 89% accuracy in video deepfake detection
  • Custom models trained by the research team
  • Dual-pipeline approach for comprehensive media verification

Technologies

PythonPyTorchComputer VisionAudio ProcessingDeep Learning

Outcome

End-to-end deepfake detection pipeline with experiment tracking and model evaluation.

Deepfake Generation Platform

End-to-end platform for generating high-quality synthetic media using cutting-edge AI technologies for ethical research purposes.

Key Highlights

  • Integration with Spanish F5 for voice synthesis
  • FaceFusion for realistic face swapping
  • Sora2 integration for video generation
  • HeyGen for AI avatar creation
  • Built for ethical research and detection model training

Technologies

Spanish F5FaceFusionSora2HeyGenPython

Outcome

Synthetic media generation platform for research and detection model training.

FitnessAI

AI-powered fitness application using computer vision for exercise form correction and personalized workout recommendations. Available on web and iOS.

Key Highlights

  • Real-time pose estimation and form analysis
  • ML-based personalized workout generation
  • Cross-platform: Web application and iOS native app
  • iOS app: github.com/AlexReinosoPerez/iOSminiAppFitness

Technologies

PythonMediaPipeTensorFlowFlaskOpenCVSwiftiOS

Outcome

Functional AI-driven fitness application with real-time pose analysis.