Regression-Guided Refocusing Learning with Feature Alignment for Remote Sensing Tiny Object Detection

计算机科学 目标检测 人工智能 特征(语言学) 遥感 特征提取 计算机视觉 模式识别(心理学) 对象(语法) 地质学 语言学 哲学
作者
Ling Ge,Guanqun Wang,Tong Zhang,Yin Zhuang,He Chen,Hao Dong,Liang Chen
出处
期刊:IEEE Transactions on Geoscience and Remote Sensing [Institute of Electrical and Electronics Engineers]
卷期号:: 1-1
标识
DOI:10.1109/tgrs.2024.3407122
摘要

Tiny object detection is a formidable challenge in remote sensing intelligent interpretation. Tiny objects are usually fuzzy, densely distributed and highly sensitive to positioning errors, which leads to the mainstream detector usually achieving suboptimal detection performance when facing tiny objects. To address the mismatch of mainstream detector architectures and model optimization strategies in the context of tiny object detection, this paper presents an efficient and interpretable algorithm for tiny object detection, termed the Cross-Attention based Feature Fusion Enhanced tiny object detection Network (CAF 2 ENet). First, the cross-attention mechanism is introduced to refine the upsampling results of deep features. This refinement improves the precision of multi-scale feature fusion. Second, a training strategy named regression-based refocusing learning is introduced. Deviating from the conventional optimization strategy, our method guides the optimizer to prioritize higher-quality detection boxes by adjusting sample weights. This adjustment significantly amplifies the detector's potential to achieve superior detection results. Finally, the object composite confidence score is employed for the interpretable filtering of detection boxes. Extensive experiments on Tiny Object Detection in Aerial Images (AI-TOD) and object Detection in Optical Remote sensing images (DIOR) datasets are carried out, and comparison indicate that the proposed CAF 2 ENet can perform the remarkable performance compared to other state-of-the-art (SOTA) tiny object detection detectors, as it can reach 63.7% Average Precision ( AP 50 ) on AI-TOD and 75.4% AP 50 on DIOR, achieve SOTA performance.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
PDF的下载单位、IP信息已删除 (2025-6-4)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
刚刚
杨扬完成签到,获得积分0
刚刚
susu完成签到,获得积分10
刚刚
詹雅智完成签到 ,获得积分10
刚刚
迅速谷云发布了新的文献求助10
刚刚
完美世界应助典雅的俊驰采纳,获得10
刚刚
英俊的铭应助关显锋采纳,获得10
2秒前
2秒前
2秒前
Xie完成签到,获得积分10
3秒前
飞飞发布了新的文献求助10
3秒前
LLC完成签到,获得积分10
5秒前
cc驳回了乐乐应助
5秒前
5秒前
junfeiwang完成签到,获得积分10
6秒前
量子星尘发布了新的文献求助10
7秒前
YanHv1发布了新的文献求助10
7秒前
7秒前
汉堡包应助JohnZhao采纳,获得10
7秒前
weeqe完成签到,获得积分10
7秒前
kecheng完成签到,获得积分10
8秒前
果酱君完成签到,获得积分10
8秒前
9秒前
JamesPei应助过过过采纳,获得10
9秒前
2以李完成签到,获得积分10
10秒前
10秒前
Akim应助森森采纳,获得10
10秒前
南有乔木完成签到,获得积分10
10秒前
充电宝应助典雅的俊驰采纳,获得10
11秒前
汉堡包应助无情的凌文采纳,获得10
11秒前
闪闪寒云完成签到 ,获得积分10
11秒前
张有志完成签到,获得积分10
11秒前
12秒前
12秒前
jane发布了新的文献求助10
12秒前
温柔寒梅完成签到 ,获得积分10
12秒前
拼搏绿柳完成签到,获得积分10
13秒前
jiajin完成签到,获得积分20
14秒前
最佳损友完成签到,获得积分0
14秒前
YanHv1完成签到,获得积分10
14秒前
高分求助中
A new approach to the extrapolation of accelerated life test data 1000
Cognitive Neuroscience: The Biology of the Mind 1000
Technical Brochure TB 814: LPIT applications in HV gas insulated switchgear 1000
ACSM’s Guidelines for Exercise Testing and Prescription, 12th edition 500
Nucleophilic substitution in azasydnone-modified dinitroanisoles 500
不知道标题是什么 500
A Preliminary Study on Correlation Between Independent Components of Facial Thermal Images and Subjective Assessment of Chronic Stress 500
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 遗传学 基因 物理化学 催化作用 冶金 细胞生物学 免疫学
热门帖子
关注 科研通微信公众号,转发送积分 3969033
求助须知:如何正确求助?哪些是违规求助? 3513900
关于积分的说明 11170818
捐赠科研通 3249256
什么是DOI,文献DOI怎么找? 1794708
邀请新用户注册赠送积分活动 875326
科研通“疑难数据库(出版商)”最低求助积分说明 804759