鱼眼图像处理论文《FisheyeMODNet: Moving Object detection on Surround-view Cameras for Autonomous Driving》阅读感悟

自己略读了一下，网上资源也不多，有读相同论文的小伙伴，欢迎补充。

一、解决的问题：

Moving Object Detection (MOD)

a fisheye surround-view system that captures a 360 view of the scene

that were captured in autonomous driving environment.

二、技术方案

1、we will make an improved version of the current dataset public to encourage further research.

2、To target embedded deployment, we design a lightweight encoder sharing weights across sequential images.

 

                                                                                         网络结构

三、创新点

1、Generation of the first public automotive dataset for fisheye images with MOD annotations.

2、 Implementation of an efficient two-stream network architecture suitable for embedded systems.

3、 Empirical study of different training and data augmentation schemes.

四、实验方案与结果

五、可能存在的问题

1、鱼眼图像与普通图像不一样，训练时，能否根据成像原理，对损失函数进行优化？

（In future work, we plan to incorporate geometric priors into the loss function to improve accuracy.）

暂时就到这里，欢迎大家补充讨论。在另一篇文章中，将会鱼眼图像公布数据集 WoodScape