logo
CanyamAI summaries for academic research
language
MobileMobile
  • Phone FeatureWeChat Mini Program
EnglishEnglish
noticeNotice
BlogBlog
delete
arrow
Return

VALA: Virtual anchor-guided label assignment for tiny object detection

delete2026-05-21
delete0
PRE
AI
J
Jiyun Zhang
Q
Qiang Fang *
S
Shuohao Shi
Y
Yujun Zeng
徐鑫 cover
徐鑫 (Xin Xu)
DOI：10.1016/j.neucom.2026.134055delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
• VALA is proposed to address anchor–object scale mismatch in tiny object detection. • Virtual IoU (VIoU) is introduced to provide scale-aware similarity measurement for tiny objects. • A Dynamic Scaling Strategy is designed to improve training stability and positive sample selection. • VALA can be seamlessly integrated into existing anchor-based detectors without architectural changes. • VALA achieves consistent improvements on multiple tiny and general object detection benchmarks.
Keywords:
VALA
tiny object detection
virtual IoU
label assignment
anchor-based detection

Journal

Neurocomputing cover
Neurocomputing
IF:
6.5
Papers:
2.5W
Citations:
6.5W

Organization

No organization information available