代码地址：Gitee、Github
存储地址：
百度云-提取码：
Google云
- DSSD-Model(pytorch版本)
《DSSD:Deconvolutional Single Shot Detector》
—基于DSSD的目标检测
作者：Cheng-Yang ,Wei Liu
单位：UNC Chapel、 Amazon Inc
发表会议及时间：CVPR 2017
Submission history
[v1] Mon, 23 Jan 2017 22:33:35 UTC (6,659 KB)
- 补充：可以尝试通过增加上下文内容(context)，来提高mAP
本文就在反卷积部分增加了context,将高层特征和底层特征相融合
- Abstract
The main contribution of this paper is an approach for introducing additional context into state-of-the-art general object detection. To achieve this we first combine a state-of-the-art classifier (Residual-101[14]) with a fast detection framework (SSD[18]).
We then augment SSD+Residual-101 with deconvolution layers to introduce additional large-scale context in object detection and improve accuracy, especially for small objects, calling our resulting system DSSD for deconvolutional single shot detector. While these two contributions are easily described at a high-level, a naive implementation does not succeed.
Instead we show that carefully adding additional stages of learned transformations, specifically a module for feed-forward connections in deconvolution and a new output module, enables this new approach and forms a potential way forward for further detection research.
Results are shown on both PASCAL VOC and COCO detection. Our DSSD with 513×513 input achieves 81.5% mAP on VOC2007 test, 80.0% mAP on VOC2012 test, and 33.2% mAP on COCO, outperforming a state-of-the-art method R-FCN[3] on each dataset
一 论文导读
DSSD是SSD算法改进分支中最为著名的一个，SSD其他有名的改进：RSSD、FSSD等
DSSD改进了SSD对小目标不够鲁棒的缺点。
主要改进点：
- 提出基于Top down的网络结构，用反卷积代替传统的双线性插值上采样
- 在预测阶段引入残差单元，优化候选框回归和分类任务输入的特征图
- 采用两阶段训练方法
DSSD在VOC2007数据集上，513*513的输入图片上，DSSD的mAP值达到了81.5%，但是fps只有6fps左右
所以它的缺点就是速度慢
补充：
- 双线性插值法
- 反卷积
- 比较
二 论文精读
三 代码实现
四 问题思索
