Research and industrial anomaly detection Peer-reviewed
GRD-Net
Generative-reconstructive-discriminative anomaly detection with region-of-interest attention.
PyTorchComputer visionAnomaly detectionIndustrial inspection
Research, engineering, deployment
Niccolò Ferrari
Senior Machine Learning Engineer and Researcher, Ph.D.
I build computer vision and machine learning systems for industrial inspection, from research prototypes to C++ and Python runtimes that have to work on real machines.
NF.π
vision.anomaly.edge.llm Now
Working where models meet cameras, hardware, latency, and maintenance.
Industrial computer visionVisual anomaly detectionEdge AI and embedded inferenceProduction ML systemsLocal agents and LLM toolingC++ and Unreal EngineEmbedded firmware and hardware
Featured
Selected projects
Research and industrial deployment Under review
DeepGen Industrial Integration
Integration of deep generative anomaly detection into a high-speed industrial line.
Deep generative modelsIndustrial lineComputer visionDeployment
Industrial computer vision and anomaly detection Preprint
MH-PatchCore
Covariance-aware and streaming-compatible PatchCore research for industrial anomaly detection.
PatchCoreMahalanobis distancePyTorchEdge AIAnomaly detection
LLMs, local agents, and orchestration Preprint
Graph Memory Transformer (GMT)
Co-authored language model architecture research around graph-structured memory.
TransformersLanguage modelsGraph memoryPyTorch
Research
Published work, thesis research, and current experiments
2023 Peer-reviewed / Open access
GRD-Net: Generative-Reconstructive-Discriminative Anomaly Detection with Region of Interest Attention Module
Peer-reviewed work on generative, reconstructive, and discriminative visual anomaly detection for industrial inspection.
2026 Preprint / Under review
Integration of Deep Generative Anomaly Detection Algorithm in High-Speed Industrial Line
Research on bringing deep generative anomaly detection into a high-speed industrial inspection line.
2026 Preprint / Submitted to EAAI
Mahalanobis PatchCore: Covariance-Aware and Streaming-Compatible Industrial Anomaly Detection
arXiv preprint on covariance-aware, streaming-compatible PatchCore for industrial anomaly detection, submitted to EAAI.