Arturo Ghinassi's Projects
Binary and Multi-class Image Segmentation of High Resolution Aerial Drone Images
Implementation of an API in fastapi connected to a MongoDB server and hosted via Docker.
Real World Video Super Resolution Architectures to super-resolute highly corrupted and noisy images
Topic Modeling with BERT using as baseline Dimensionality Reduction and Clustering on the TF-IDF Matrix.
Vision Transformer Architectures for Image Denoising
Cosine Annealing scheduler with linear warmup and support for multiple parameters groups.
Implementing the Linkage and Pathway Universal Tests using the clrdag package for Direceted Acyclic Graphs. Testing the two tests on cytometry data to analyze the links.
Docker container for a loan prediction application.
A fast MoE impl for PyTorch
CounTR applied on FITS Images for the task of Object Counting
An all MLP Architecture for Image Super-Resolution
Simple tutorial for the FastMoE library.
Multilevel Beta regression to extrapolate MotoGP riderβs skill accounting for constructor advantage.
Analysis of the lyrics evolution from the 50s till today.
system unit to push ngrok url every time it changes.
RAFT unofficial implementation in PyTorch
TL;DR about me
Implementation of a Simple Siamese network for contrastive learning in tensorflow2.0
Video Super Resolution library that implements training and evaluation of video super resolution architectures. Metrics and SoA architectures are already implemented, for quick fine-tuning and evaluation.