Multi-Prototype Guided Source-Free Domain Adaptive Object Detection for Autonomous Driving

计算机科学 正规化(语言学) 一致性(知识库) 人工智能 质心 领域(数学分析) 班级(哲学) 探测器 对象(语法) 分割 模式识别(心理学) 目标检测 机器学习 数学 电信 数学分析
作者
Siqi Zhang,Lu Zhang,Guangsen Li,Pengcheng Li,Zhiyong Liu
出处
期刊:IEEE transactions on intelligent vehicles [Institute of Electrical and Electronics Engineers]
卷期号:9 (1): 1589-1601
标识
DOI:10.1109/tiv.2023.3337795
摘要

Source-free domain adaptive object detection (source-free DAOD) seeks to adapt a detector pre-trained on a source domain to an unlabeled target domain without requiring access to annotated source domain data. To address challenges posed by domain shifts, current source-free DAOD approaches mainly rely on the self-training paradigm, where pseudo labels are predicted and employed to fine-tune the detector on unlabeled target domain. However, these methods often encounter issues related to intra-class variation, resulting in category-specific biases and noisy pseudo labels. In response, we present an effective Multi-Prototype Guided source-free DAOD method, dubbed MPG, consisting of two key components: multi-prototype guided pseudo labeling (MPPL) and multi-prototype guided consistency regularization (MPCR) modules. In the MPPL module, we construct category-specific multiple prototypes to better represent the category with intra-class variations. Specifically, multiple prototypes with adaptive cluster centroids are introduced for each category to effectively capture the intra-class variations. Through the implementation of the proposed MPPL module, we derive more accurate pseudo labels by assessing the proximity of instance features to multiple category prototypes. In the MPCR module, we introduce multi-level consistency regularization, including prototype-based consistency and prediction consistency, which encourages the model to overlook style perturbations and learn domain-invariant representations. Extensive experiments on five public driving datasets demonstrate that MPG outperforms existing state-of-the-art methods, showcasing its effectiveness in adapting object detectors to target domains.
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
1秒前
喜悦的水云完成签到 ,获得积分10
1秒前
钱念波完成签到,获得积分10
2秒前
逍遥自在完成签到,获得积分10
3秒前
倪小呆完成签到 ,获得积分10
3秒前
5秒前
山神厘子完成签到,获得积分10
5秒前
娇娇大王完成签到,获得积分10
6秒前
Zpear应助qhjqljqd采纳,获得10
7秒前
8秒前
10秒前
xiaxia42完成签到 ,获得积分10
11秒前
小蘑菇应助DIY101采纳,获得10
12秒前
Ashley完成签到 ,获得积分10
13秒前
Hydrogen发布了新的文献求助10
13秒前
Bioflying完成签到,获得积分10
13秒前
15秒前
爆米花应助无限的以亦采纳,获得10
16秒前
曾宪俊完成签到 ,获得积分10
17秒前
QZZ完成签到,获得积分10
18秒前
18秒前
离子电池完成签到,获得积分10
18秒前
留胡子的路灯完成签到,获得积分10
20秒前
silin完成签到,获得积分10
21秒前
传奇3应助zzf采纳,获得10
21秒前
qhjqljqd完成签到,获得积分10
22秒前
踏月偷心发布了新的文献求助10
22秒前
roy_chiang完成签到,获得积分10
22秒前
DIY101发布了新的文献求助10
23秒前
Hydrogen完成签到,获得积分10
23秒前
dong完成签到 ,获得积分10
25秒前
满鑫完成签到,获得积分10
26秒前
贺丞完成签到,获得积分10
26秒前
东风完成签到,获得积分10
26秒前
杨一完成签到 ,获得积分10
27秒前
chen完成签到,获得积分10
27秒前
朴实寻真完成签到,获得积分10
28秒前
火星上白羊完成签到,获得积分10
30秒前
阳光的幻雪完成签到 ,获得积分10
31秒前
Lucas应助LLLLLLLL采纳,获得10
32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
Immigrant Incorporation in East Asian Democracies 600
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 3968578
求助须知:如何正确求助?哪些是违规求助? 3513406
关于积分的说明 11167631
捐赠科研通 3248853
什么是DOI,文献DOI怎么找? 1794499
邀请新用户注册赠送积分活动 875150
科研通“疑难数据库(出版商)”最低求助积分说明 804671