2021 CVPR-ICCV等目标检测

1、You Only Look One-level Feature
分析fpn作用，减小fpn大小，提高模型速度
2、Dynamic Head: Unifying Object Detection Heads with Attentions
3、Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection
从预测框的分数入手，不好训练，容易崩
4、 PSRR-MaxpoolNMS: Pyramid Shifted MaxpoolNMS with Relationship Recovery
5、IQDet: Instance-wise Quality Distribution Sampling for Object Detection
6、Adaptive Class Suppression Loss for Long-Tail Object Detection
7、VarifocalNet: An IoU-aware Dense Object Detector
从预测框的分数入手，（不好训练，容易崩）
8、OPANAS: One-Shot Path Aggregation Network Architecture Search for Object Detection
9、Multiple Instance Active Learning for Object Detection
10、Equalization Loss v2: A New Gradient Balance Approach for Long-Tailed Object Detection
参考：https://mp.weixin.qq.com/s/x2TuAxesAb3dfJr-Cyzpsg
https://github.com/amusi/CVPR2021-Papers-with-Code#Object-Detection

ICCV
TOOD

其他：
YOLOX

版权声明：本文为CSDN博主「yang_daxia」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/yang_daxia/article/details/119212823