用少量人工标注提升视觉模型精度，靠的是智能筛选标签质量。
Fine-tuning Vision Classifiers On A Budget
- 基于标签者可信度构建贝叶斯模型，推断真实标签
- 在相同预算下，标注数据量提升30%以上，模型性能不变
- 适合资源有限但需高精度标注的工业视觉场景
微调现代计算机视觉模型需要准确标注的数据，但真实标签可能不存在，而可通过不同准确率的标注者获得多组标签。本文将标签质量与标注者可信度关联，证明在已知标注者准确率的前提下，使用简单的朴素贝叶斯模型估算真实标签，可在固定预算下增加标注数据量而不降低标签或微调质量。我们在工业图像数据集上进行了实验，结果表明，所提出的“真实标签扩展”（Ground Truth Extension, GTX）方法能以更少的人工标注完成模型微调。
原文摘要 · Abstract (English)
Fine-tuning modern computer vision models requires accurately labeled data for which the ground truth may not exist, but a set of multiple labels can be obtained from labelers of variable accuracy. We tie the notion of label quality to confidence in labeler accuracy and show that, when prior estimates of labeler accuracy are available, using a simple naive-Bayes model to estimate the true labels allows us to label more data on a fixed budget without compromising label or fine-tuning quality. We present experiments on a dataset of industrial images that demonstrates that our method, called Ground Truth Extension (GTX), enables fine-tuning ML models using fewer human labels.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。