Name: Danilo Đorđević
Type: User
Company: ETH Zürich
Bio: MSc graduate @ ETH Zürich, researching the broad area of deep learning, part. transformers, information retrieval, GNNs, generative models, both in CV and NLP.
Location: Zurich, Switzerland
Danilo Đorđević's Projects
Python notebook containing to some selected leetcode/hackerrank problems.
The modification of the original "Fine-tuning CNN Image Retrieval with No Human Annotation" paper, where I use Vision Transformer architectures instead of CNNs to extract features.
Implementation of the original denoising diffusion probabilistic model (DDPM) paper in PyTorch.
This repository contains code used to perform image retrieval using transformers. It is a demonstration of how using vision transformers, metric learning, and a novel loss based on differential cross-entropy can lead to better retrieval than classical CNN-based methods.
[CVPR 2024 - Oral, Best Paper Award Candidate] Marigold: Repurposing Diffusion-Based Image Generators for Monocular Depth Estimation
This repository contains my original code solutions and project reports, as well as the provided problem formulations for assignments 1, 2, 3 and 6 for the Natural Language Processing course, held in the Autumn semester 2022 at ETH Zürich. Each folder contains the files of the corresponding assignment.
Code solutions for assignements in the Probabilistic Artificial Intelligence course in the Autumn semester 2021 at ETH Zurich.
Implementations of RL algorithms, starting from MDPs.