Box-Level Class-Balanced Sampling for Active Object Detection
This work addresses the annotation burden in object detection for researchers and practitioners, but it is incremental as it builds on existing box-level active learning methods.
The paper tackles the problem of expensive bounding box annotation in object detection by proposing a class-balanced sampling strategy for box-level active learning, which selects more minority class objects for labeling and uses task-aware soft pseudo labeling to improve pseudo label accuracy, achieving state-of-the-art performance on public datasets.
Training deep object detectors demands expensive bounding box annotation. Active learning (AL) is a promising technique to alleviate the annotation burden. Performing AL at box-level for object detection, i.e., selecting the most informative boxes to label and supplementing the sparsely-labelled image with pseudo labels, has been shown to be more cost-effective than selecting and labelling the entire image. In box-level AL for object detection, we observe that models at early stage can only perform well on majority classes, making the pseudo labels severely class-imbalanced. We propose a class-balanced sampling strategy to select more objects from minority classes for labelling, so as to make the final training data, \ie, ground truth labels obtained by AL and pseudo labels, more class-balanced to train a better model. We also propose a task-aware soft pseudo labelling strategy to increase the accuracy of pseudo labels. We evaluate our method on public benchmarking datasets and show that our method achieves state-of-the-art performance.