On Model Calibration for Long-Tailed Object Detection and Instance Segmentation

Abstract

Vanilla models for object detection and instance segmentation suffer from the heavy bias toward detecting frequent objects in the long-tailed setting. Existing methods address this issue mostly during training, e.g., by re-sampling or reweighting. In this paper, we investigate a largely overlooked approach -postprocessing calibration of confidence scores. We propose NORCAL, Normalized Calibration for long-tailed object detection and instance segmentation, a simple and straightforward recipe that reweighs the predicted scores of each class by its training sample size. We show that separately handling the background class and normalizing the scores over classes for each proposal are keys to achieving superior performance. On the LVIS dataset, NORCAL can effectively improve nearly all the baseline models not only on rare classes but also on common and frequent classes. Finally, we conduct extensive analysis and ablation studies to offer insights into various modeling choices and mechanisms of our approach. * Equal contributions Preprint. Under review.

Relieving Long-tailed Instance Segmentation via Pairwise Class Balance

“…Similarly, ACSL [38] only penalize negative classes over threshold. Separating the categories into some small groups [21,42] and simple calibration [26,47] helps, too. [28,36] modified the original soft-max function by embedding the distribution prior, achieving success.…”

“…Comparison with NORCAL[26] which is also complementary to other methods. NORCAL is grid searched and reported the results from optimal hyper-parameters for each method.…”