Behzad Bozorgtabar

Computer Vision Group Leader & Lecturer at EPFL
Senior Scientist at CHUV-EPFL

email: behzad.bozorgtabar [at] epfl [dot] ch

RESEARCH UPDATES

CVPR 2023 paper: 'CrOC: Cross-View Online Clustering for Dense Visual Representation Learning'

CrOC

CVPR 2023 paper: 'TeSLA: Test-Time Self-Learning With Automatic Adversarial Augmentation'

TeSLA

About Me

I'm a senior scientist at the Signal Processing Lab (LTS5) at the Swiss Federal Institute of Technology (EPFL), with a joint affiliation with the Lausanne University Hospital (CHUV) Department of Radiology, Lausanne, Switzerland. At the EPFL-LTS5, I am the computer vision leader for the medical imaging group. I am also a European Lab for Learning & Intelligent Systems (ELLIS) member. I am also a research staff scientist at the Center for Biomedical Imaging (CIBM). Earlier, I was a Postdoctoral Researcher at IBM Research-Australia.

My principal research area lies at the intersection of computer vision and medical image analysis using machine learning techniques. I have a strong interest in domain adaptation/generalization and self-supervised learning on learning from limited data or labels, which I consider major avenues for innovation and impact for many vision-based applications. My research's ultimate goal is to develop robust deep image representations that capture and understand the world, as well as our human eye and mind, do.

News

[Feb 2023] Our paper, TeSLA: Test-Time Self-Learning With Automatic Adversarial Augmentation, has received acceptance from CVPR 2023.
[Feb 2023] Our paper, CrOC: Cross-View Online Clustering for Dense Visual Representation Learning, has received acceptance from CVPR 2023.
[Jan 2023] Our joint work with the University of Bern, namely 'SAGTTA' on test-time augmentation, has been accepted to ISBI 2023.
[Nov 2022] Our paper on self-supervised anomaly localization has received acceptance from AAAI 2023 (acceptance rate of 19.6%) (Oral).
[Oct 2022] Our paper won First Runner up award at the AIMIA Workshop at ECCV 2022.
[Oct 2022] Our new paper, ScoreNet, has received acceptance from WACV 2023.
[Sep 2022] Our new paper on transformer-based anomaly detection and localization has received acceptance from BMVC 2022.

EPFL Computer Vision Talks

EPFL CV Talks. I am organizing the EPFL Computer Vision Talks
YouTube Channel

Citations

citations. A complete list of my publications and patents can be found at Google Scholar

Recent Selected Publications

CrOC_main *New* CrOC: Cross-View Online Clustering for Dense Visual Representation Learning
Thomas Stegmüller, Tim Lebailly, Behzad Bozorgtabar, Tinne Tuytelaars, Jean-Philippe Thiran
CVPR 2023
project page · paper · github

TeSLA_main *New* TeSLA: Test-Time Self-Learning With Automatic Adversarial Augmentation
Devavrat Tomar, Guillaume Vray, Behzad Bozorgtabar, Jean-Philippe Thiran
CVPR 2023
project page · paper · github

*New* ScoreNet: Learning Non-Uniform Attention and Augmentation for
Transformer-Based Histopathological Image Classification
Thomas Stegmüller, Behzad Bozorgtabar, Antoine Spahr, Jean-Philippe Thiran
WACV 2023
project page · paper

ANOM Anomaly Detection and Localization Using Attention-Guided Synthetic Anomaly
and Test-Time Adaptation
Behzad Bozorgtabar, Dwarikanath Mahapatra, Jean-Philippe Thiran
BMVC 2022
paper

OptTTA OptTTA: Learnable Test-Time Augmentation for
Source-Free Medical Image Segmentation Under Domain Shift
Devavrat Tomar, Guillaume Vray, Jean-Philippe Thiran, Behzad Bozorgtabar
MIDL 2022 (Oral)
project page · paper · github

SRMA. Self-Rule to Multi-Adapt: Generalized Multi-source Feature Learning
Using Unsupervised Domain Adaptation for Colorectal Cancer Tissue Detection
Christian Abbet, Linda Studer, Andreas Fischer, Heather Dawson, Inti Zlobec,
Behzad Bozorgtabar, Jean-Philippe Thiran
MedIA Journal 2022
paper · github

SST. Self-Supervised Generative Style Transfer for One-Shot Medical Image Segmentation
Devavrat Tomar, Behzad Bozorgtabar, Manana Lortkipanidze, Guillaume Vray,
Mohammad Saeed Rad, Jean-Philippe Thiran
WACV 2022
paper · github

SegGini. Learning Whole-Slide Segmentation from Inexact and
Incomplete Labels using Tissue Graphs
Valentin Anklin, Pushpak Pati, Guillaume Jaume, Behzad Bozorgtabar,
Antonio Foncubierta-Rodríguez, Jean-Philippe Thiran, Mathilde Sibony,
Maria Gabrani, Orcun Goksel
MICCAI 2021
paper · github

SOoD. SOoD: Self-Supervised Out-of-Distribution Detection
Under Domain Shift for Multi-Class Colorectal Cancer Tissue Types
Behzad Bozorgtabar, Guillaume Vray, Dwarikanath Mahapatra, Jean-Philippe Thiran
ICCVW 2021
paper · github

CVPR2021. Quantifying Explainers of Graph Neural Networks in Computational Pathology
Guillaume Jaume, Pushpak Pati, Behzad Bozorgtabar, Antonio Foncubierta-Rodríguez,
Florinda Feroce, Anna Maria Anniciello, Tilman Rau, Maria Gabrani,
Jean-Philippe Thiran, Orcun Goksel
CVPR 2021
paper · github

T-MI. Self-Attentive Spatial Adaptive Normalization for Cross-Modality Domain Adaptation
Devavrat Tomar, Manana Lortkipanidze, Guillaume Vray,
Behzad Bozorgtabar, Jean-Philippe Thiran
IEEE T-MI 2021
paper · github

MIDL2021. Self-Rule to Adapt: Learning Generalized Features from Sparsely-Labeled Data
Using Unsupervised Domain Adaptation for Colorectal Cancer Tissue Phenotyping
Christian Abbet, Linda Studer, Andreas Fischer, Heather Dawson, Inti Zlobec,
Behzad Bozorgtabar, Jean-Philippe Thiran
MIDL 2021
paper · github

ISBI2021. Self-Taught Semi-Supervised Anomaly Detection on Upper Limb X-rays
Antoine Spahr , Behzad Bozorgtabar, Jean-Philippe Thiran
ISBI 2021
paper· github

ISBI2021. Benefiting from Bicubically Down-Sampled Images for
Learning Real-World Image Super-Resolution
Mohammad Saeed Rad, Thomas Yu, Claudiu Musat, Hazım Kemal Ekenel,
Behzad Bozorgtabar, Jean-Philippe Thiran
WACV 2021
paper

SALAD. SALAD: Self-Supervised Aggregation Learning for Anomaly Detection on X-Rays
Behzad Bozorgtabar, Dwarikanath Mahapatra, Guillaume Vray, Jean-Philippe Thiran
MICCAI 2020
paper · github

Self-Rule. Divide-and-Rule: Self- Supervised Learning for Survival Analysis in Colorectal Cancer
Christian Abbet, Inti Zlobec, Behzad Bozorgtabar, Jean-Philippe Thiran
MICCAI 2020
paper · github

CVPR2020. Pathological Retinal Region Segmentation From OCT Images
Using Geometric Relation Based Augmentation
Dwarikanath Mahapatra, Behzad Bozorgtabar, Jean-Philippe Thiran, Ling Shao
CVPR 2020
paper

Syndemo. SynDeMo: Synergistic Deep Feature Alignment for Joint Learning of Depth and Ego-Motion
Behzad Bozorgtabar, Mohammad Saeed Rad, Dwarikanath Mahapatra, Jean-Philippe Thiran
ICCV 2019
paper · supplementary material

SROBB. SROBB: Targeted Perceptual Loss for Single Image Super-Resolution
Mohammad Saeed Rad, Behzad Bozorgtabar, Urs-Viktor Marti,
Max Basler, Hazım Kemal Ekenel,
Jean-Philippe Thiran
ICCV 2019
paper · supplementary material

FG2019. Using Photorealistic Face Synthesis and Domain Adaptation
to Improve Facial Expression Analysis
Behzad Bozorgtabar, Mohammad Saeed Rad, Hazım Kemal Ekenel, Jean-Philippe Thiran
FG 2019
paper · github

CVIU2019. Learn to Synthesize and Synthesize to Learn
Behzad Bozorgtabar, Mohammad Saeed Rad, Hazım Kemal Ekenel, Jean-Philippe Thiran
CVIU 2019
paper · github

NeurIPS18. Image-Level Attentional Context Modeling Using Nested-Graph Neural Networks
Guillaume Jaume, Behzad Bozorgtabar, Hazım Kemal Ekenel,
Jean-Philippe Thiran, Maria Gabrani
NeurIPS 2018
paper

MSMCT. MSMCT: Multi-State Multi-Camera Tracker
Behzad Bozorgtabar, Roland Goecke
IEEE TCSVT 2018
paper

Grants

Personalized Health and Related Technologies (PHRT)

Swiss Cancer League

Discovery Translation Fund (DTF 2.0)

Teaching

2019-Present	Image analysis and pattern recognition (EE-451-4 ECTS- Bozorgtabar & Thiran), EPFL
2019-Present	Lab in signal and image processing (EE-490(f)-4 ECTS- Bozorgtabar & Thiran), EPFL