5 papers
Probabilistic Label Spreading: Efficient and Consistent Estimation of Soft Labels with Epistemic Uncertainty on Graphs
Jonathan Klees, Tobias Riedlinger, Peter Stehr +3
Safe artificial intelligence for perception tasks remains a major challenge, partly due to the lack of data with high-quality labels. Annotations themselves are subject to aleatori…
From Label Error Detection to Correction: A Modular Framework and Benchmark for Object Detection Datasets
Sarina Penquitt, Jonathan Klees, Rinor Cakaj +3
Object detection has advanced rapidly in recent years, driven by increasingly large and diverse datasets. However, label errors often compromise the quality of these datasets and a…
Quantifying Ambiguity in Categorical Annotations: A Measure and Statistical Inference Framework
Christopher Klugmann, Daniel Kondermann
Human-generated categorical annotations frequently produce empirical response distributions (soft labels) that reflect ambiguity rather than simple annotator error. We introduce an…
Decomposing Visual Classification: Assessing Tree-Based Reasoning in VLMs
Sary Elmansoury, Islam Mesabah, Gerrit GroÃmann +4
Vision language models (VLMs) excel at zero-shot visual classification, but their performance on fine-grained tasks and large hierarchical label spaces is understudied. This paper…
Minority Reports: Balancing Cost and Quality in Ground Truth Data Annotation
Hsuan Wei Liao, Christopher Klugmann, Daniel Kondermann +1
High-quality data annotation is an essential but laborious and costly aspect of developing machine learning-based software. We explore the inherent tradeoff between annotation accu…