5 citations · 18 across the 15 of their papers we have counts for
15 papers
Locating Demographic Bias at the Attention-Head Level in CLIP's Vision Encoder
Alaa Yasser, Kittipat Phunjanna, Marcos Escudero Viñolo +2
Standard fairness audits of foundation models quantify that a model is biased, but not where inside the network the bias resides. We propose a mechanistic fairness audit that combi…
Leveraging Contrastive Learning for Semantic Segmentation with Consistent Labels Across Varying Appearances
Javier Montalvo, Roberto Alcover-Couso, Pablo Carballeira +3
This paper introduces a novel synthetic dataset that captures urban scenes under a variety of weather conditions, providing pixel-perfect, ground-truth-aligned images to facilitate…
VLMs meet UDA: Boosting Transferability of Open Vocabulary Segmentation with Unsupervised Domain Adaptation
Roberto Alcover-Couso, Marcos Escudero-Viñolo, Juan C. SanMiguel +1
Segmentation models are typically constrained by the categories defined during training. To address this, researchers have explored two independent approaches: adapting Vision-Lang…
Pinpoint Counterfactuals: Reducing social bias in foundation models via localized counterfactual generation
Kirill Sirotkin, Marcos Escudero-Viñolo, Pablo Carballeira +3
Foundation models trained on web-scraped datasets propagate societal biases to downstream tasks. While counterfactual generation enables bias analysis, existing methods introduce a…
Layer-wise Model Merging for Unsupervised Domain Adaptation in Segmentation Tasks
Roberto Alcover-Couso, Juan C. SanMiguel, Marcos Escudero-Viñolo +1
Merging parameters of multiple models has resurfaced as an effective strategy to enhance task performance and robustness, but prior work is limited by the high costs of ensemble cr…
Gradient-based Class Weighting for Unsupervised Domain Adaptation in Dense Prediction Visual Tasks
Roberto Alcover-Couso, Marcos Escudero-Viñolo, Juan C. SanMiguel +1
In unsupervised domain adaptation (UDA), where models are trained on source data (e.g., synthetic) and adapted to target data (e.g., real-world) without target annotations, address…