Saved in:
Bibliographic Details
Main Authors: Neubauer, Melanie, Rueckert, Elmar, Rauch, Christian
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.09701
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913022161715200
author Neubauer, Melanie
Rueckert, Elmar
Rauch, Christian
author_facet Neubauer, Melanie
Rueckert, Elmar
Rauch, Christian
contents Detecting unseen anomalies in unstructured environments presents a critical challenge for industrial and agricultural applications such as material recycling and weeding. Existing perception systems frequently fail to satisfy the strict operational requirements of these domains, specifically real-time processing, pixel-level segmentation precision, and robust accuracy, due to their reliance on exhaustively annotated datasets. To address these limitations, we propose a weakly supervised pipeline for object segmentation and classification using weak image-level supervision called 'Patch Aggregation for Segmentation of Targets and Anomalies' (PASTA). By comparing an observed scene with a nominal reference, PASTA identifies Target and Anomaly objects through distribution analysis in self-supervised Vision Transformer (ViT) feature spaces. Our pipeline utilizes semantic text-prompts via the Segment Anything Model 3 to guide zero-shot object segmentation. Evaluations on a custom steel scrap recycling dataset and a plant dataset demonstrate a 75.8% training time reduction of our approach to domain-specific baselines. While being domain-agnostic, our method achieves superior Target (up to 88.3% IoU) and Anomaly (up to 63.5% IoU) segmentation performance in the industrial and agricultural domain.
format Preprint
id arxiv_https___arxiv_org_abs_2604_09701
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle PASTA: Vision Transformer Patch Aggregation for Weakly Supervised Target and Anomaly Segmentation
Neubauer, Melanie
Rueckert, Elmar
Rauch, Christian
Computer Vision and Pattern Recognition
Machine Learning
Detecting unseen anomalies in unstructured environments presents a critical challenge for industrial and agricultural applications such as material recycling and weeding. Existing perception systems frequently fail to satisfy the strict operational requirements of these domains, specifically real-time processing, pixel-level segmentation precision, and robust accuracy, due to their reliance on exhaustively annotated datasets. To address these limitations, we propose a weakly supervised pipeline for object segmentation and classification using weak image-level supervision called 'Patch Aggregation for Segmentation of Targets and Anomalies' (PASTA). By comparing an observed scene with a nominal reference, PASTA identifies Target and Anomaly objects through distribution analysis in self-supervised Vision Transformer (ViT) feature spaces. Our pipeline utilizes semantic text-prompts via the Segment Anything Model 3 to guide zero-shot object segmentation. Evaluations on a custom steel scrap recycling dataset and a plant dataset demonstrate a 75.8% training time reduction of our approach to domain-specific baselines. While being domain-agnostic, our method achieves superior Target (up to 88.3% IoU) and Anomaly (up to 63.5% IoU) segmentation performance in the industrial and agricultural domain.
title PASTA: Vision Transformer Patch Aggregation for Weakly Supervised Target and Anomaly Segmentation
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2604.09701