[Paper note] Learning from Simulated and Unsupervised Images through Adversarial Training
来源:互联网 发布:网络喷子 编辑:程序博客网 时间:2024/04/30 04:33
- paper
- This is (probably?) the first paper from Apple in ML/CV field
Contribution
- Propose Simulate + Unsupervised training, using real world images to refine the synthetic images, with GAN
- Training GAN using adversarial loss and a self-regularization loss
- Key modifications to stabilize GAN training and prevent artifacts
Framework
- Refiner (Generator):
x̃ :=Rθ(x) - Refiner loss (general formular):
R(θ)=∑ilreal(θ;x̃ i,)+λlreg(θ;x̃ i,xi) lreg minimizing the difference between the synthetic and the refined images
- Discriminator Loss:
D(ϕ)=−∑ilog(Dϕ(x̃ i))−∑jlog(1−Dϕ(yj)) x~_i, y_j
are randomly sampled from refined images and real images sets
- Algorithm:
L_R
loss in the implementation of this paper:R(θ)=∑ilog(1−Dϕ(Rθ(xi)))+λ||Rθ(xi)−xi||1
Stabilize GAN training
- Local adversarial loss: divide the refined image and real image into
w x h
regions and use separate discriminators to judge each region- Final loss is the sum of loss on each region
- Using history of refined images
- Two issues when only use the latest refined images
- Diverging of adversarial training
- The refiner network re-introducing the artifacts that the discriminator had forgotten about
- Buffer history images, use
b/2
history images andb/2
newly refined images in each iter - Update
b/2
of the buffered images in each iter
- Two issues when only use the latest refined images
Experiments
- Gaze estimation
- Dataset: MPIIGaze dataset
- Synthesizer: UnityEyes
- Visual Turing test: human cannot tell the difference between refined and real images
- Quantitative result: 22.3% percentage of improvement
- Hand pose estimation
- Dataset: NYU hand pose dataset
- Training CNN: Hour glass network
0 0
- [Paper note] Learning from Simulated and Unsupervised Images through Adversarial Training
- 每日论文 Learning from Simulated and Unsupervised Images through Adversarial Training
- 对抗学习之Learning from Simulated and Unsupervised Images through Adversarial Training
- [paper]Unsupervised learning of disparity maps from stereo images
- 【Paper Note】Convolutional Clustering for Unsupervised Learning 论文翻译
- 【Paper Note】Convolutional Clustering for Unsupervised Learning 论文理解
- 挖坑:UNSUPERVISED AND SEMI-SUPERVISED LEARNING WITH CATEGORICAL GENERATIVE ADVERSARIAL NETWORKS
- 《Unsupervised Learning of Depth and Ego-Motion from Video》读书笔记
- 【Paper Note】Generative Adversarial Nets 论文理解
- Unsupervised feature learning by augmenting single images
- Unsupervised Rrpresentation Learning with Deep Convolutional Generative Adversarial Networks
- 【Paper Note】Generative Adversarial Nets论文翻译(未完待续)
- ML_1 supervised learning and unsupervised learning
- CS231n学习笔记--16. Adversarial Examples and Adversarial Training
- Paper:Learning from Imbalanced Data
- [读PAPER]Online Unsupervised Feature Learning for Visual Tracking
- [Paper note] Joint Learning of Single-image and Cross-image Representations for Person Re-id.
- Unsupervise learning——Unsupervised feature learning and deep learning
- 三大主流开源 NoSQL 数据库和两大主流传统 SQL 数据库对比
- Caffe中Flages的安装与使用!
- Big Data Ecosystem and Components
- (39.1) Spring Boot Shiro权限管理【从零开始学Spring Boot】
- 07、一步一步学thinkjs之实现注销以及判断是否登录
- [Paper note] Learning from Simulated and Unsupervised Images through Adversarial Training
- 路径规划
- STL排序算法之swap()
- 第一个微信小程序demo
- Linux之守护进程
- maven 镜像地址
- css使用li奇数行或偶数行
- linux 下makefile 编译 Android
- 初学servlet,从servlet往jsp页面发送list集合,报错