Publication

Which Private Attributes do VLMs Agree on and Predict Well?

Andrea Cavallaro
2026

Abstract

Visual Language Models (VLMs) are often used for zero-shot detection of visual attributes in the image. We present a zero-shot evaluation of open-source VLMs for privacy-related attribute recognition. We identify the attributes for which VLMs exhibit strong inter-annotator agreement, and discuss the disagreement cases of human and VLM annotations. Our results show that when evaluated against human annotations, VLMs tend to predict the presence of privacy attributes more often than human annotators. In addition to this, we find that in cases of high inter-annotator agreement between VLMs, they can complement human annotation by identifying attributes overlooked by human annotators. This highlights the potential of VLMs to support privacy annotations in large-scale image datasets.

Official source

https://infoscience.epfl.ch/entities/publication/45aa3d16-91ed-43a7-9505-c7150b4cbdfe

About this result

This page is automatically generated and may contain information that is not correct, complete, up-to-date, or relevant to your search query. The same applies to every other page on this website. Please make sure to verify the information with EPFL's official sources.

Graph Chat

Chat with Graph Search

Ask any question about EPFL courses, lectures, exercises, research, news, etc. or try the example questions below.

DISCLAIMER: The Graph Chatbot is not programmed to provide explicit or categorical answers to your questions. Rather, it transforms your questions into API requests that are distributed across the various IT services officially administered by EPFL. Its purpose is solely to collect and recommend relevant references to content that you can explore to help you answer your questions.

Andrea Cavallaro
2026

Abstract

Official source

https://infoscience.epfl.ch/entities/publication/45aa3d16-91ed-43a7-9505-c7150b4cbdfe

About this result

Related concepts (19)

Human Genome Project

The Human Genome Project (HGP) was an international scientific research project with the goal of determining the base pairs that make up human DNA, and of identifying, mapping and sequencing all of the genes of the human genome from both a physical and a functional standpoint. It started in 1990 and was completed in 2003. It remains the world's largest collaborative biological project. Planning for the project started after it was adopted in 1984 by the US government, and it officially launched in 1990.

Large language model

A large language model (LLM) is a language model characterized by its large size. Their size is enabled by AI accelerators, which are able to process vast amounts of text data, mostly scraped from the Internet. The artificial neural networks which are built can contain from tens of millions and up to billions of weights and are (pre-)trained using self-supervised learning and semi-supervised learning. Transformer architecture contributed to faster training.

Human genome

The human genome is a complete set of nucleic acid sequences for humans, encoded as DNA within the 23 chromosome pairs in cell nuclei and in a small DNA molecule found within individual mitochondria. These are usually treated separately as the nuclear genome and the mitochondrial genome. Human genomes include both protein-coding DNA sequences and various types of DNA that does not encode proteins. The latter is a diverse category that includes DNA coding for non-translated RNA, such as that for ribosomal RNA, transfer RNA, ribozymes, small nuclear RNAs, and several types of regulatory RNAs.