Fernanda Gonçalves Abrantes

arXiv2025

The Timepix4 Beam Telescope

The spatial and temporal performance of a telescope system comprising planar silicon sensors bump-bonded to Timepix4-v2 ASICs are assessed at the CERN SPS using a $180$~GeV/$c$ mixed hadron beam. The pointing resolution at the centre of the telescope is $2.3 \pm 0.1\;\mathrm{μm}$ and $2.4 \pm 0.1\mathrm{μm}$ in x and y directions, respectively. The temporal resolution for the combined timestamps on a track using only the information from the Timepix4 planes is found to be $92 \pm 5$~ps. The telescope is extensively described including its data acquisition system and dedicated software as well as the corrections applied to the raw measurements.

hep-ex↗

arXiv2023

VisoGender: A dataset for benchmarking gender bias in image-text pronoun resolution

We introduce VisoGender, a novel dataset for benchmarking gender bias in vision-language models. We focus on occupation-related biases within a hegemonic system of binary gender, inspired by Winograd and Winogender schemas, where each image is associated with a caption containing a pronoun relationship of subjects and objects in the scene. VisoGender is balanced by gender representation in professional roles, supporting bias evaluation in two ways: i) resolution bias, where we evaluate the difference between pronoun resolution accuracies for image subjects with gender presentations perceived as masculine versus feminine by human annotators and ii) retrieval bias, where we compare ratios of professionals perceived to have masculine and feminine gender presentations retrieved for a gender-neutral search query. We benchmark several state-of-the-art vision-language models and find that they demonstrate bias in resolving binary gender in complex scenes. While the direction and magnitude of gender bias depends on the task and the model being evaluated, captioning models are generally less biased than Vision-Language Encoders. Dataset and code are available at https://github.com/oxai/visogender

cs.CV↗