Solution for Large-Scale Hierarchical Object Detection Datasets with Incomplete Annotation and Data Imbalance
This is an incremental improvement for computer vision researchers and practitioners working on object detection in complex, real-world datasets.
The paper tackled the problem of large-scale hierarchical object detection with incomplete annotations and data imbalance in the Open Images 2018 Challenge, achieving a mAP of 62.2% on the public leaderboard and 58.6% on the private leaderboard, ranking 2nd and 3rd respectively.
This report demonstrates our solution for the Open Images 2018 Challenge. Based on our detailed analysis on the Open Images Datasets (OID), it is found that there are four typical features: large-scale, hierarchical tag system, severe annotation incompleteness and data imbalance. Considering these characteristics, an amount of strategies are employed, including SNIPER, soft sampling, class-aware sampling (CAS), hierarchical non-maximum suppression (HNMS) and so on. In virtue of these effective strategies, and further using the powerful SENet154 armed with feature pyramid module and deformable ROIalign as the backbone, our best single model could achieve a mAP of 56.9%. After a further ensemble with 9 models, the final mAP is boosted to 62.2% in the public leaderboard (ranked the 2nd place) and 58.6% in the private leaderboard (ranked the 3rd place, slightly inferior to the 1st place by only 0.04 point).