[2005.08455] Large-Scale Object Detection in the Wild from Imbalanced Multi-Labelsopen searchopen navigation menucontact arXivarXiv Twitter

Training with more data has always been the most stable and effective way of improving performance in deep learning era. As the largest object detection dataset so far, Open Images brings great opportunities and challenges for object detection in general and sophisticated scenarios. However, owing to its semi-automatic collecting and labeling pipeline to deal with the huge data scale, Open Images dataset suffers from label-related problems that objects may explicitly or implicitly have multiple labels and the label distribution is extremely imbalanced. In this work, we quantitatively analyze these label problems and provide a simple but effective solution. We design a concurrent softmax to handle the multi-label problems in object detection and propose a soft-sampling methods with hybrid training scheduler to deal with the label imbalance. Overall, our method yields a dramatic improvement of 3.34 points, leading to the best single model with 60.90 mAP on the public object detection tes

2 mentions: @kuritateppei@arxiv_cs_cv_pr
Date: 2020/05/20 03:51

Referring Tweets

@kuritateppei Open Imagesなどのマルチラベルが多い大規模データセットに対して効果的な改良型Softmax Lossの提案。マルチラベルの場合、Softmax Lossに直接複数の正解ラベルを割り当てると各ラベルのスコアが互いに抑制されてしまうのを防ぐように自然な拡張を実現。CVPR2020オーラル。 t.co/NnyTNHICPL t.co/sLv8rSODA6

Related Entries

Read more [2004.05746] Enabling Incremental Knowledge Transfer for Object Detection at the Edgeopen searchopen...
0 users, 2 mentions 2020/04/20 06:52
Read more [2004.14899] Polarization Human Shape and Pose Datasetopen searchopen navigation menucontact arXivar...
0 users, 2 mentions 2020/05/04 03:51
Read more [2005.03860] Where am I looking at? Joint Location and Orientation Estimation by Cross-View Matching...
0 users, 2 mentions 2020/05/12 02:21
Read more [2005.05659] Stillleben: Realistic Scene Synthesis for Deep Learning in Roboticsopen searchopen navi...
0 users, 2 mentions 2020/05/16 02:21
Read more [2005.09623] Focus on defocus: bridging the synthetic to real domain gap for depth estimationopen se...
0 users, 2 mentions 2020/05/21 00:52