This is my blog.
沉迷于为什么我的3D框这么混乱（就有很多个很多个），特别是有多余的框，但是学长的就那么正常，那么干净。干净了许多，虽然还有一些奇怪的框，不过毕竟切入角度不一样，释然了
深度估计
深度估计字面意思而已。
同样从人来讲的话，是一件很简单的事情。对于眼睛捕捉的一张图像，大脑自然而然解析出的是一个三维关系，而对于网络呢，他可能需要根据整个图像的信息、物体周围信息、透视法、物体本身的大小形状位置等信息，或者在图像序列中会加上时序关系，亦或者相机的变动根据焦距的关系等等判断出这个物体在整张图中的相对深度（其实很多时候一直在做的就是探索大脑是怎么做的，也不完全是）。
现在我看到的更多的是单目深度估计，可能是我检索有些问题，也可能是钱不够用，所以在技术上探索？单目深度估计的问题即MDE，Monocular Depth Estimation。单目问题在Godard e t c., 2017中指出，所谓单目是指在测试的时候只用了一张图片。一般来说，单目数据会是一个序列，因此也有很多方法是利用了时间的关系。所以在训练的时候可以用两个摄像机的图像来训练咯？人来说的话都是双目深度估计吧，双眼之间是有距离的，然后根据几何关系感觉就可以；但是人闭上一只眼睛，却也可以较准确地给出深度关系，这么一看刚出生的我就很机智哎
self-supervised learning before 2019
This is my blog.
Self-Supervised Learning has become an exciting direction in AI community.
- Jitendra Malik: “Supervision is the opium of the AI researcher”
- Alyosha Efros: “The AI revolution will not be supervised”
- Yann LeCun: “self-supervised learning is the cake, supervised learning is the icing on the cake, reinforcement learning is the cherry on the cake”
So, this blog is about the classic algorithm in self-supervised learning before 2019.
(Not finished)
self-supervised learning in latest
This is my blog.
Self-Supervised Learning has become an exciting direction in AI community.
- Jitendra Malik: “Supervision is the opium of the AI researcher”
- Alyosha Efros: “The AI revolution will not be supervised”
- Yann LeCun: “self-supervised learning is the cake, supervised learning is the icing on the cake, reinforcement learning is the cherry on the cake”
So, this blog is about the recent improvement (including STA) in self-supervised learning (for researchers).
self-supervised learning-introduction
This is my blog.
Self-Supervised Learning has become an exciting direction in AI community.
- Jitendra Malik: “Supervision is the opium of the AI researcher”
- Alyosha Efros: “The AI revolution will not be supervised”
- Yann LeCun: “self-supervised learning is the cake, supervised learning is the icing on the cake, reinforcement learning is the cherry on the cake”
So, this blog is about the introduction about self-supervised learning (only for beginners, like me).