Transferring Labels to Solve Annotation Mismatches Across Object Detection Datasets

Yuan-Hong Liao, David Acuna, Rafid Mahmood, James Lucas, Viraj Uday Prabhu, Sanja Fidler

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: object detection, data-centric AI, label translation, dataset improvements

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

TL;DR: We introduce the label translation problem for object detection, where we modify the annotations of a labeled source dataset to match the label protocols in a target dataset.

Abstract: In object detection, varying annotation protocols across datasets can result in annotation mismatches, leading to inconsistent class labels and bounding regions. Addressing these mismatches typically involves manually identifying common trends and fixing the corresponding bounding boxes and class labels. To alleviate this laborious process, we introduce the label transfer problem in object detection. Here, the goal is to transfer bounding boxes from one or more source datasets to match the annotation style of a target dataset. We propose a data-centric approach, Label-Guided Pseudo-Labeling (LGPL), that improves downstream detectors in a manner agnostic to the detector learning algorithms and model architectures. Validating across four object detection scenarios, defined over seven different datasets and three different architectures, we show that transferring labels for a target task via LGPL consistently improves the downstream detection in every setting, on average by $1.88$ mAP and 2.65 AP$^{75}$. Most importantly, we find that when training with multiple labeled datasets, carefully addressing annotation mismatches with LGPL alone can improve downstream object detection better than off-the-shelf supervised domain adaptation techniques that align instance features.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Primary Area: transfer learning, meta learning, and lifelong learning

Submission Number: 334