Library of autoencoders for sequential data

Geometric Adversarial Attacks and Defenses on 3D Point Clouds (3DV 2021)

SAGA: Spectral Adversarial Geometric Attack on 3D Meshes (ICCV 2023)

A text to audio pipeline using Riffusion (a finetuned stablediffusion model) and using RAVE a audio to audio AutoEncoder.

Richter's Predictor: Modeling Earthquake Damage Challenge 0.7521 Scored Solution

CelebFaces is my computer vision gym, where I solve different tasks, test different approaches and learn a lot.

➕💓Let's build the Simplest Possible Autoencoder . ⁉️🏷We'll start Simple, with a Single fully-connected Neural Layer as Encoder and as Decoder. 👨🏻‍💻🌟An Autoencoder is a type of Artificial Neural Network used to Learn Efficient Data Codings in an unsupervised manner🌘🔑

Proposing a structure for a convolutional neural network capable of coloring grayscale images. The study focused on images of streets within cities. The generative neural network was trained on as many street images as possible.