Fine-tuning Vision Classifiers On A Budget
Keywords: fine-tuning, label efficiency, crowd sourcing, learning with noisy labels, label aggregation, active learning, computer vision
TL;DR: We show how to aggregate noisy labels more efficiently for better fine-tuning with fewer human labels.
Abstract: 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 label quality to confidence derived from historical labeler accuracy using a simple naive-Bayes model. Imputing true labels in this way 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.
Submission Number: 54
Loading