5 papers · 1 filter
Pruning by Block Benefit: Exploring the Properties of Vision Transformer Blocks during Domain Adaptation
Patrick Glandorf, Bodo Rosenhahn
Vision Transformer have set new benchmarks in several tasks, but these models come with the lack of high computational costs which makes them impractical for resource limited hardw…
Multi-Flow: Multi-View-Enriched Normalizing Flows for Industrial Anomaly Detection
Mathis Kruse, Bodo Rosenhahn
With more well-performing anomaly detection methods proposed, many of the single-view tasks have been solved to a relatively good degree. However, real-world production scenarios o…
HydraMix: Multi-Image Feature Mixing for Small Data Image Classification
Christoph Reinders, Frederik Schubert, Bodo Rosenhahn
Training deep neural networks requires datasets with a large number of annotated examples. The collection and annotation of these datasets is not only extremely expensive but also…
Utilizing Uncertainty in 2D Pose Detectors for Probabilistic 3D Human Mesh Recovery
Tom Wehrbein, Marco Rudolph, Bodo Rosenhahn +1
Monocular 3D human pose and shape estimation is an inherently ill-posed problem due to depth ambiguities, occlusions, and truncations. Recent probabilistic approaches learn a distr…
SplatPose & Detect: Pose-Agnostic 3D Anomaly Detection
Mathis Kruse, Marco Rudolph, Dominik Woiwode +1
Detecting anomalies in images has become a well-explored problem in both academia and industry. State-of-the-art algorithms are able to detect defects in increasingly difficult set…