Multiple object stitching for unsupervised representation learning

Chengchao Shen

doi:10.1016/j.patcog.2025.112602

Multiple object stitching for unsupervised representation learning

2025-10-15

0

PRE

AI

DOI：10.1016/j.patcog.2025.112602

Original

Original request for help

Save

Abstract

En 中文

• Single-object contrastive methods suffer inferior performance on multi-object images. • We propose a novel multiple object stitching to learn multi-object representations. • Our method constructs delicate multi-object correspondences to learn multi-object representations. • Our method achieves superior performance on various downstream tasks.

Multiple object stitching for unsupervised representation learning

Abstract

Journal

Organization