Publications

A brief summary of my publications, also available on Google Scholar.
 

DINOv2 Meets Text: A Unified Framework for Image- and Pixel-Level Vision-Language Alignment
CVPR 2025 - Locked-image tuning for vision-language alignment using a DINOv2 backbone and a few tricks on top.
Cijo Jose, Théo Moutakanni, Dahyun Kang, Federico Baldassarre, Timothée Darcet, Hu Xu, Daniel Li, Marc Szafraniec, Michaël Ramamonjisoa, Maxime Oquab, Oriane Siméoni, Huy v. Vo, Patrick Labatut, Piotr Bojanowski
PDF
DINOv2 Meets Text: A Unified Framework for Image- and Pixel-Level Vision-Language Alignment
Cluster and Predict Latents Patches for Improved Masked Image Modeling
TMLR 2025 - Stable training of dense image representations using a clustering loss on ViT patch tokens.
Timothée Darcet, Federico Baldassarre, Maxime Oquab, Julien Mairal, Piotr Bojanowski
PDF Code
Cluster and Predict Latents Patches for Improved Masked Image Modeling
Variable Rate Allocation for Vector-Quantized Autoencoders
ICASSP 2023 - Control the bit rate allocation of a a PQ-VAE using semantic information from DINO, in collaboration with Meta AI in Paris.
Federico Baldassarre, Alaaeldin El-Nouby, Hervé Jégou
PDF Cite Poster
Variable Rate Allocation for Vector-Quantized Autoencoders
Quantitative Metrics for Evaluating Explanations of Video DeepFake Detectors
BMVC 2022 - Quantitative metrics for evaluating explanations of video DeepFake detectors, in collaboration with the Huawei Ireland Research Center.
Federico Baldassarre, Quentin Debard, Gonzalo Fiz Pontiveros, Tri Kurniawan Wijaya
PDF Cite Code Poster Video DOI
Quantitative Metrics for Evaluating Explanations of Video DeepFake Detectors
Learnable Masked Tokens for Improved Transferability of Self-Supervised Vision Transformers
ECML PKDD 2022 - Learnable masked tokens reduce overfitting when training transformers in low-data regime.
Hao Hu, Federico Baldassarre, Hossein Azizpour
PDF Cite
Learnable Masked Tokens for Improved Transferability of Self-Supervised Vision Transformers
Towards Self-Supervised Learning of Global and Object-Centric Representations
ICLR Workshop 2022 - How to learn object-centric representations using contrastive learning on multi-object datasets?
Federico Baldassarre, Hossein Azizpour
PDF Cite Code Poster DOI
Towards Self-Supervised Learning of Global and Object-Centric Representations
Modeling SARS-CoV-2 proteins in the CASP-commons experiment
Proteins: Structure, Function, and Bioinformatics - CASP community-wide experiment on modeling SARS-CoV-2 proteins causing the COVID-19 pandemic.
Andriy Kryshtafovych, John Moult, Wendy Billings, Dennis Della Corte, Krzysztof Fidelis, Sohee Kwon, Kliment Olechnovič, Chaok Seok, Česlovas Venclovas, Jonghun Won, CASP-COVID Participants
PDF Cite DOI
Modeling SARS-CoV-2 proteins in the CASP-commons experiment
Explanation-based Weakly-supervised Learning of Visual Relations with Graph Networks
ECCV 2020 - A novel weakly-supervised method for visual relationship detection that relies on minimal image-level predicate labels, graph networks, and attribution-based explanations.
Federico Baldassarre, Kevin Smith, Josephine Sullivan, Hossein Azizpour
PDF Cite Code Video DOI
Explanation-based Weakly-supervised Learning of Visual Relations with Graph Networks
GraphQA: Protein Model Quality Assessment Using Graph Convolutional Networks
Bioinformatics 2020 - A novel method for protein quality assessment based on a graphical representation of tertiary structures and graph convolutional networks.
Federico Baldassarre, David Menéndez Hurtado, Arne Elofsson, Hossein Azizpour
PDF Cite Code DOI
GraphQA: Protein Model Quality Assessment Using Graph Convolutional Networks
Explainability Techniques for Graph Convolutional Networks
ICML Workshop 2019 - How do Sensitivity Analysis, Guided Backpropagation, and Layer-Wise Relevance Propagation behave when applied to Graph Neural Networks?
Federico Baldassarre, Hossein Azizpour
PDF Cite Code Poster DOI
Explainability Techniques for Graph Convolutional Networks
Morphing Architectures for Pose-based Image Generation of People in Clothing
Master’s Thesis - Given the picture of a person and of a clothing item, generate the image of the person wearing it. Work performed during an internship at Zalando Research, Berlin.
Federico Baldassarre
PDF Cite
Morphing Architectures for Pose-based Image Generation of People in Clothing
Deep Koalarization
Technical Report - Grayscale image colorization using a CNN conditione on Inception-ResNet-v2. Deep Learning course at KTH, best project award.
Federico Baldassarre, Diego Gonzalez Morín, Lucas Rodés-Guirao
PDF Cite Code DOI
Deep Koalarization