[论文解读] SIoU Loss: More Powerful Learning for Bounding Box Regression

Zhora Gevorgyan|arXiv (Cornell University)|May 25, 2022

Advanced Neural Network Applications被引用 563

一句话总结

SIoU 引入一种用于边界框回归的损失，考虑预测框与真实框之间的不匹配方向，从而提升训练速度和准确性。

ABSTRACT

The effectiveness of Object Detection, one of the central problems in computer vision tasks, highly depends on the definition of the loss function - a measure of how accurately your ML model can predict the expected outcome. Conventional object detection loss functions depend on aggregation of metrics of bounding box regression such as the distance, overlap area and aspect ratio of the predicted and ground truth boxes (i.e. GIoU, CIoU, ICIoU etc). However, none of the methods proposed and used to date considers the direction of the mismatch between the desired ground box and the predicted, "experimental" box. This shortage results in slower and less effective convergence as the predicted box can "wander around" during the training process and eventually end up producing a worse model. In this paper a new loss function SIoU was suggested, where penalty metrics were redefined considering the angle of the vector between the desired regression. Applied to conventional Neural Networks and datasets it is shown that SIoU improves both the speed of training and the accuracy of the inference. The effectiveness of the proposed loss function was revealed in a number of simulations and tests.

研究动机与目标

• 推动改进目标检测中的边界框回归损失函数。
• 解决现有损失忽略框之间不匹配方向的局限性。
• 提出通过预测框与真实框之间的夹角重新定义惩罚项的 SIoU 损失。
• 在标准数据集上展示训练速度和推断精度的提升。

提出的方法

• 定义包含预测框到真实框向量角度的惩罚项的 SIoU 损失。
• 重新表述距离、重叠和纵横比的考虑，以包括方向信息。
• 将 SIoU 集成到标准的目标检测神经网络中，并在常见数据集上进行评估。
• 与传统损失（如 GIoU、CIoU、ICIoU）相比，比较收敛速度和准确性。

实验结果

研究问题

• RQ1 在回归损失中加入方向信息是否能加速训练过程的收敛？
• RQ2 与现有损失相比，SIoU 能否提升边界框回归的准确性？
• RQ3 SIoU 如何影响典型目标检测流水线中的推断性能？

主要发现

• SIoU 相对于传统损失提高了训练效率。
• 在实证测试中，SIoU 提高了边界框预测的准确性。
• 方向性惩罚项在优化过程中使收敛更快。
• 该方法与传统的神经网络检测器和数据集兼容。