Supervision Interpolation via LossMix: Generalizing Mixup for Object Detection and Beyond
This addresses the problem of enhancing object detector performance and robustness for computer vision researchers, representing a novel method rather than an incremental improvement.
The paper tackled the challenge of applying data mixing augmentations to object detection by introducing Supervision Interpolation and LossMix, which interpolates loss errors instead of ground truth labels, resulting in consistent performance improvements on PASCAL VOC and MS COCO datasets and setting a new state of the art in cross-domain detection.
The success of data mixing augmentations in image classification tasks has been well-received. However, these techniques cannot be readily applied to object detection due to challenges such as spatial misalignment, foreground/background distinction, and plurality of instances. To tackle these issues, we first introduce a novel conceptual framework called Supervision Interpolation (SI), which offers a fresh perspective on interpolation-based augmentations by relaxing and generalizing Mixup. Based on SI, we propose LossMix, a simple yet versatile and effective regularization that enhances the performance and robustness of object detectors and more. Our key insight is that we can effectively regularize the training on mixed data by interpolating their loss errors instead of ground truth labels. Empirical results on the PASCAL VOC and MS COCO datasets demonstrate that LossMix can consistently outperform state-of-the-art methods widely adopted for detection. Furthermore, by jointly leveraging LossMix with unsupervised domain adaptation, we successfully improve existing approaches and set a new state of the art for cross-domain object detection.