Classification Calibration for Long-tail Instance Segmentation
This addresses the problem of long-tail instance segmentation for computer vision applications, but it is incremental as it builds on existing two-stage models with a calibration approach.
The paper tackles the performance drop of state-of-the-art instance segmentation models on long-tail data, finding inaccurate classification as a major cause, and proposes a calibration method that improves tail class performance by a large margin without significant architectural changes.
Remarkable progress has been made in object instance detection and segmentation in recent years. However, existing state-of-the-art methods are mostly evaluated with fairly balanced and class-limited benchmarks, such as Microsoft COCO dataset [8]. In this report, we investigate the performance drop phenomenon of state-of-the-art two-stage instance segmentation models when processing extreme long-tail training data based on the LVIS [5] dataset, and find a major cause is the inaccurate classification of object proposals. Based on this observation, we propose to calibrate the prediction of classification head to improve recognition performance for the tail classes. Without much additional cost and modification of the detection model architecture, our calibration method improves the performance of the baseline by a large margin on the tail classes. Codes will be available. Importantly, after the submission, we find significant improvement can be further achieved by modifying the calibration head, which we will update later.