RetinaNet算法的简介(论文介绍)
RetinaNet源自论文Focal Loss for Dense Object Detection用于密集目标检测的焦损失。
Abstract
The highest accuracy object detectors to date are based on a two-stage approach popularized by R-CNN, where a classifier is applied to a sparse set of candidate object locations. In contrast, one-stage detectors that are applied over a regular, dense sampling of possible object locations have the potential to be faster and simpler, but have trailed the accuracy of two-stage detectors thus far. In this paper, we investigate why this is the case. We discover that the extreme foreground-background class imbalance encountered during training of dense detectors is the central cause. We propose to address this class imbalance by reshaping the standard cross entropy loss such that it down-weights the loss assigned to well-classified examples. Our novel Focal Loss focuses training on a sparse set of hard examples and prevents the vast number of easy negatives from overwhelming the detector during training. To evaluate the effectiveness of our loss, we design and train a simple dense detector we call RetinaNet. Our results show that when trained with the focal loss, RetinaNet is able to match the speed of previous one-stage detectors while surpassing the accuracy of all existing state-of-the-art two-stage detectors. Code is at: https://github.com/facebookresearch/Detectron.
迄今为止，精度最高的目标检测器是基于R-CNN推广的两阶段方法，其中分类器应用于稀疏的一组候选对象位置。相比之下，对可能的目标位置进行常规、密集采样的单级探测器有可能更快、更简单，但迄今仍落后于两级探测器的精度。在本文中，我们将探讨为什么会出现这种情况。我们发现，在密集探测器训练过程中所遇到的极端的前-后级不平衡是其主要原因。我们建议通过重新构造标准的交叉熵损失来解决这个类的不平衡，这样它就可以降低分配给分类良好的示例的损失。我们的新焦失聚焦训练集中在一组稀疏的硬例子上，防止大量的容易的负片在训练中压倒检测器。为了评估我们的损失的有效性，我们设计并训练了一个简单的高密度探测器，我们称之为RetinaNet。我们的结果表明，当使用聚焦损失训练时，视黄连能够在超过现有所有最先进的两级探测器精度的同时，与以前的单级探测器速度相匹配。代码如下:https://github.com/facebookresearch/Detectron。
Conclusion
In this work, we identify class imbalance as the primary obstacle preventing one-stage object detectors from surpassing top-performing, two-stage methods. To address this, we propose the focal loss which applies a modulating term to the cross entropy loss in order to focus learning on hard negative examples. Our approach is simple and highly effective. We demonstrate its efficacy by designing a fully convolutional one-stage detector and report extensive experimental analysis showing that it achieves stateof-the-art accuracy and speed. Source code is available at https://github.com/facebookresearch/Detectron [12].
在这项工作中，我们确定类不平衡是阻止单级对象检测器超越性能最好的两级方法的主要障碍。为了解决这个问题，我们提出了焦损失，它应用一个调制项的交叉熵损失，以便集中学习硬的负面例子。我们的方法简单而高效。我们设计了一种全卷积单级检测器来验证其有效性，并报告了大量的实验分析，结果表明它达到了最先进的精度和速度。源代码可从https://github.com/facebookresearch/Detectron[12]获得。
论文
Tsung-Yi Lin, Priya Goyal, Ross Girshick, KaimingHe, Piotr Dollár.
Focal Loss for Dense Object Detection. ICCV 2017.
https://arxiv.org/abs/1708.02002
0、实验结果
1、RetinaNet-50 VS RetinaNet-101
在COCO数据集上，作者提出的RetinaNet-101-800模型，可以达到37.8mAP可超过其他以前提出的单阶段检测方法，并且检测速度是198ms。
2、在COCO test-dev上——分别与单阶段、两阶段模型比较
目标检测单模型结果（边界框AP），与COCO test-dev的最新技术相比。RetinaNet可以达到40.8mAP，甚至可以超过两阶段方法。
注：该表展示了RetinaNet-101-800模型的结果，该模型经过尺度抖动训练。模型取得了最佳成绩，超越了其它单阶段和两阶段模型。
1、RetinaNet算法的特点及其贡献
找到妨碍单阶段目标检测器实现高准确度的主要原因：
-训练期间的前景-背景之间的类别不平衡
设计焦点损失(Focal Loss) 来解决这种类别不平衡问题，降低分配给分类良好例子的损失。
提出了单阶段RetinaNet网络架构，使用了焦点损失和多尺度特征金字塔。
RetinaNet算法的架构详解
DL之RetinaNet：RetinaNet算法的架构详解https://yunyaniu.blog.csdn.net/article/details/100010853
RetinaNet算法的案例应用
DL之RetinaNet：基于RetinaNet算法(keras框架)利用resnet50_coco数据集(.h5文件)实现目标检测
https://blog.csdn.net/qq_41185868/article/details/100109252
