Lecture 2 | Image Classification

作者: Ysgc | 来源:发表于2019-11-03 12:40 被阅读0次

semantic gap: huge gap between a "cat" and these matrices

completely different matrices, but they all represent the same cat

only part of a cat is recorded

explicitly rule -> fail -> 1. not stable, 2. not transferrable to other classes -> can't scale up

rotation invariant -> L2

decision boundary

L2 is bad at describing the perceptual distance

图像分类
Lecture 2: Image Classification pipeline Image Classifica...
Lecture 2 | Image Classification
semantic gap: huge gap between a "cat" and these matrices
Lecture2： Image Classification P
Lecture2 Image Classification Pipeline 1. 图片在计算机中的储存方式&图像...
lecture 6
Continue from lecture 5: tagging as classification, combi...
#cs231n#1.最近邻分配器Nearest Neighbor
The Image Classification Pipeline The image classificatio...
tensorflow 2.x系列02 图像分类
tensorflow 2系列02 图像分类(Image classification using TensorFl...
Image Classification
Nearest Neighbor Classifier L1 vs. L2. It is interesting ...
Image classification
一、首先是一种非训练的方式直接比较后进行归类： Nearest Neighbor:比较方式包括：L1距离，L2距离...
Lecture 4 Classification
我们再来举宝可梦的例子：比如说宝可梦有很多属性，现在input一个皮卡丘，那么output就是雷。那么问题来了，...
Lecture 4 - Classification