An unsupervised pipeline for class-agnostic object detection using self-supervised vision transformers and Kolmogorov–Arnold Networks

Object detection typically requires costly manual bounding box annotations. We propose a fully unsupervised four-stage pipeline that automatically generates pseudo-bounding boxes: (1) Background Removal, (2) Pseudo-Bounding Box Generation, (3) Complementary Bounding Box Generation (optional), and (4...

Description complète

Détails bibliographiques

Auteurs principaux:	Yuno Otsuka, Pittipol Kantavat, Boonserm Kijsirikul
Format:	Artigo
Langue:	Inglês
Publié:	Elsevier 2026-06-01
Collection:	Machine Learning with Applications
Sujets:	Class-agnostic object detection Kolmogorov–Arnold Networks Self-supervised learning Unsupervised bounding box generation Unsupervised object localization Vision transformers
Accès en ligne:	http://www.sciencedirect.com/science/article/pii/S2666827026000356
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!