[论文解读] Resolving Class Imbalance in Object Detection with Weighted Cross Entropy Losses

Trong Huy Phan, Kazuma Yamamoto|arXiv (Cornell University)|Jun 2, 2020

Advanced Neural Network Applications参考文献 19被引用 54

一句话总结

这篇论文研究加权交叉熵损失在缓解对象检测中的类别不平衡的作用，比较 Balanced Cross Entropy、Focal Loss 与 Class-Balanced Loss 在如 BDD100K 这样的不平衡数据集上的表现。

ABSTRACT

Object detection is an important task in computer vision which serves a lot of real-world applications such as autonomous driving, surveillance and robotics. Along with the rapid thrive of large-scale data, numerous state-of-the-art generalized object detectors (e.g. Faster R-CNN, YOLO, SSD) were developed in the past decade. Despite continual efforts in model modification and improvement in training strategies to boost detection accuracy, there are still limitations in performance of detectors when it comes to specialized datasets with uneven object class distributions. This originates from the common usage of Cross Entropy loss function for object classification sub-task that simply ignores the frequency of appearance of object class during training, and thus results in lower accuracies for object classes with fewer number of samples. Class-imbalance in general machine learning has been widely studied, however, little attention has been paid on the subject of object detection. In this paper, we propose to explore and overcome such problem by application of several weighted variants of Cross Entropy loss, for examples Balanced Cross Entropy, Focal Loss and Class-Balanced Loss Based on Effective Number of Samples to our object detector. Experiments with BDD100K (a highly class-imbalanced driving database acquired from on-vehicle cameras capturing mostly Car-class objects and other minority object classes such as Bus, Person and Motor) have proven better class-wise performances of detector trained with the afore-mentioned loss functions.

研究动机与目标

• 将对象检测中的类别不平衡问题及其对检测器性能的影响作为动机。
• 评估加权交叉熵变体是否能提升检测器对少数类的准确性。
• 在高度不平衡的行驶数据集（BDD100K）上证明其有效性。

提出的方法

• 将加权交叉熵变体（基于有效样本数的 Balanced Cross Entropy、Focal Loss、Class-Balanced Loss）应用于对象检测器。
• 使用这些损失在不平衡数据上训练检测器，以评估按类别的性能提升。
• 使用一个行驶数据集（BDD100K）进行实验，该数据集以 Car 为多数对象，Bus、Person、Motor 为少数类别。
• 对比不同损失变体下的检测器性能，以识别对少数类别的增益。

实验结果

研究问题

• RQ1 加权交叉熵损失能否在不平衡数据集上提升每个类别的检测准确性？
• RQ2 哪种损失变体在提升少数类别性能方面最有效？
• RQ3 这些损失是否影响整体检测器性能还是仅影响驾车数据集中的少数类别？
• RQ4 BDD100K 的结果如何验证该方法在真实世界不平衡场景中的可行性？

主要发现

• 加权交叉熵变体在少数对象类别上显示了按类别的性能提升。
• 在 BDD100K 上的实验表明对较少出现的类，如 Bus、Person、Motor 有增益，但未给出具体数字。
• 论文提供证据表明在标准架构下训练的检测器，平衡/加权策略是有帮助的。
• 在所有损失中，研究确认 Focal Loss 和 Class-Balanced Loss 在不平衡设置中是有效的选项。