---
格式版本: 2
标题: "The Blind Spot in 2D Infants' Pose Estimation:Robust Learning from Noisy Annotations"
原文链接: "https://arxiv.org/abs/2609.04009"
发布日期: "2026-09-03"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Thu, 3 Sep 2026 15:47:56 UTC (4,623 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-09-06T21:25:34+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-09-06T21:24:50+08:00"
入库时间: "2026-09-06T13:25:34.956Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=Scale-up&searchtype=all"
匹配关键词:
  - "Scale-up"
  - "performance"
  - "AI"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 5
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "论文主题为婴儿姿态估计中的噪声标签处理，与超节点/AI Rack/机柜级AI基础设施完全无关。来源为arXiv搜索入口页，仅为摘要页，无任何相关技术或商业内容。"
AI质检模型: "zj-deepseek-v4-flash"
AI质检时间: "2026-09-06T21:25:49+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 0
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 5
AI摘要: "针对婴儿2D姿态估计中标注噪声问题，研究者提出REMIND方法，利用逐关键点训练动态聚类识别噪声标注，无需噪声先验即可训练模型。"
AI摘要模型: "ali-deepseek-v4-flash"
AI摘要时间: "2026-09-06T23:44:15.225Z"
采集批次: "2026年9月6日19点27分09秒"
采集批次ID: "20260906-192709-237"
去重键: "https://arxiv.org/abs/2609.04009"
---

## Computer Science > Computer Vision and Pattern Recognition

## Title:The Blind Spot in 2D Infants' Pose Estimation:Robust Learning from Noisy Annotations

[View PDF](https://arxiv.org/pdf/2609.04009) [HTML (experimental)](https://arxiv.org/html/2609.04009v1)

> Abstract:Noisy annotations pose a significant challenge for supervised deep learning, as neural networks rely on large-scale, high-quality labeled data whose corruption can severely impair model performance. Although robustness to label noise has been extensively studied for classification tasks, it remains relatively underexplored in Pose Estimation (PE). This limitation becomes critical in clinical contexts, including neonatology, where PE of preterm infants is used to support the assessment of spontaneous motility, a key indicator of neurodevelopmental trajectories. In such settings, infants' images labeling is further hindered by visual challenges (e.g., keypoint self-occlusions, caregiver interference), making the annotation process inherently susceptible to errors. To tackle noisy annotations in PE, we introduce REliable keypoint selection via Memory of traINing Dynamics (REMIND), a clustering-based keypoint-selection strategy that exploits keypoint-wise training dynamics to identify noisy labels without assuming any prior knowledge of the noise distribution, thus enabling noise-free model training. When evaluated on the proprietary NeoPose dataset, comprising 46 videos of 46 preterm infants recorded in real clinical settings, REMIND correctly identifies noisy annotations across multiple corruption scenarios, achieving up to 93\\% Area Under the Curve (AUC) with three different PE architectures used in the relevant literature. To our knowledge, this is the first study to explicitly address label noise in preterm infants' PE, paving the way for the design of trustworthy learning-based algorithms for infants'monitoring support when data quality cannot be guaranteed.

| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| --- | --- |
| Cite as: | [arXiv:2609.04009](https://arxiv.org/abs/2609.04009) \[cs.CV\] |
|  | (or [arXiv:2609.04009v1](https://arxiv.org/abs/2609.04009v1) \[cs.CV\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2609.04009](https://doi.org/10.48550/arXiv.2609.04009) |

## Submission history

From: Emanuele Cardinale \[[view email](https://arxiv.org/show-email/2a8d22ad/2609.04009)\]  
**\[v1\]** Thu, 3 Sep 2026 15:47:56 UTC (4,623 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2609.04009) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))