Fabian Dubourvieux

Fabian Dubourvieux

Auto-generated profile, based on publicly available information

Fabian Dubourvieux

Université Paris-Saclay, CEA, List, Palaiseau, France

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 November 2021 in IEEE Access
Verified
Journal Article
Open access
FDFabian Dubourvieux
Fabian Dubourvieux
ALAngelique Loesch
Angelique Loesch
RARomaric Audigier
Romaric Audigier
SASamia Ainouz
Samia Ainouz
SCStephane Canu
Stephane Canu
Unsupervised Domain Adaptation (UDA) for re-identification (re-ID) is a challenging task: to avoid a costly annotation of additional data, it aims at transferring knowledge from a domain with annotated data to a domain of interest with only unlabeled data. Pseudo-labeling approaches have proven to be effective for UDA re-ID. However, the effectiveness of these approaches heavily depends on the choice of some hyperparameters (HP) that affect the generation of pseudo-labels by clustering. The lack of annotation in the domain of interest makes this choice non-trivial. Current approaches simply reuse the same empirical value for all adaptation tasks and regardless of the target data representation that changes through pseudo-labeling training phases. As this simplistic choice may limit their performance, we aim at addressing this issue. We propose new theoretical grounds on HP selection for clustering UDA re-ID as well as method of automatic and cyclic HP tuning for pseudo-labeling UDA clustering: HyPASS. HyPASS consists in incorporating two modules in pseudo-labeling methods: (i) HP selection based on a labeled source validation set and (ii) conditional domain alignment of feature discriminativeness to improve HP selection based on source samples. Experiments on commonly used person re-ID and vehicle re-ID datasets show that our proposed HyPASS consistently improves the best state-of-the-art methods in re-ID compared to the commonly used empirical HP setting.

4 August 2022 in Computer Vision and Image Understanding
Verified
Journal Article
Closed access
FDFabian Dubourvieux
Fabian Dubourvieux
RARomaric Audigier
Romaric Audigier
ALAngélique Loesch
Angélique Loesch
SASamia Ainouz
Samia Ainouz
SCStéphane Canu
Stéphane Canu

7 April 2026
Verified
Thesis
Closed access
FDFabian Dubourvieux
Fabian Dubourvieux
Ré-identification et adaptabilité inter-domaines: La Re-Identification (re-ID) est une tâche d'association d'observations d'une même instance. La re-ID est au centre d'un grand nombre d'applications de la vision par ordinateur. L'apprentissage automatique supervisé, a permis d'inférer des caractéristiques discriminantes de l'identités pour la re-ID, à partir d'un grand nombre d'images d'instances dont les identités ont été annotées manuellement. Or, le domaine de test diffère couramment en pratique du domaine d'entraînement, si l'on souhaite déployer le système de re-ID dans un contexte différent. Cependant, une différence des domaines causé par un changement de lieu par exemple, entraîne une chute drastique des performances de re-ID sur le domaine de test. Étant donné le coût important des annotations manuelles, et face à la nécessité pratique de résoudre ce problème de chute des performances de re-ID sur un nouveau domaine de test, cette thèse propose de s'intéresser à l'adaptabilité inter-domaines.