文章目录
- L1 loss-平均绝对误差(Mean Absolute Error——MAE)
- L2 loss-均方误差(Mean Square Error——MSE)
- Smooth L1 Loss
- MAE、MSE、Smooth L1对比
- Multi-task loss
- Balance L1 loss
- IoU Loss
- GIoU Loss
- DIoU Loss、CIoU Loss
- E-IoU Loss、Focal E-IoU Loss
- Reference
L1 loss-平均绝对误差(Mean Absolute Error——MAE)
指模型预测值f(x)和真实值y之间距离的均值,公式如下: