I. List of studies
1. Comprehensive class
(1) collected a variety of the latest and most classic literature, neural network resources list: Https://github.com/robertsdionne/neural-network-papers contains the deep learning domain classic, as well as the latest and best algorithm, If you learn this list over and over again, you have already reached the great God level.
(2) Machine learning Checklist:https://github.com/ujjwalkarn/Machine-Learning-Tutorials/blob/master/README.md Of course, it also contains a variety of deep learning resources Tutorials
2. Computer Vision
(1) Computer Vision Learning Checklist:https://github.com/kjw0612/awesome-deep-vision contains the most classic deep learning paper. If the inside of the paper have been learned once, can be described as the visual field of Master Master High Master ...
3. NLP Field
(1) Natural Language Learning Checklist:HTTPS://GITHUB.COM/KEONKIM/AWESOME-NLP from NLP rookie to master's growth path study list
(2) RNN, LSTM, etc. biased to NLP algorithm classes: Http://handong1587.github.io/deep_learning/2015/10/09/rnn-and-lstm.html looks like a good look.
(3) Blog recommendation: http://www.wildml.com/contains RNN, LSTM, attention mechanism and other tutorials, the most important is to explain the easy to understand, let me wait for rookie benefit
Second, other summary
1, the Assistant experience: http://lamda.nju.edu.cn/weixs/project/CNNTricks/CNNTricks.html, which explains the deep learning of the adjustment of the experience summary, is a classic.
2. Gradient descent Summary of various variants: http://sebastianruder.com/optimizing-gradient-descent/written by the author is very good, easy to understand
Third, training data
Human face data:
1. Training data set of the Chinese University of Hong Kong: http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html
This database contains 20w face images, each with 5 feature points, and dozens of attributes (whether smiling, skin color, hair color, gender, etc.)
2, 68 facial feature points: http://ibug.doc.ic.ac.uk/resources/facial-point-annotations/
This site can be downloaded to 3000, training data pictures, each picture marked 68 facial feature points
3, 74 facial feature points: http://gaps-zju.org/DDE/
This web site has a 1.4w face training data picture, each picture marked 74 facial feature points. But this data feels very imprecise.
4. Gender and Age training data: https://data.vision.ee.ethz.ch/cvl/rrothe/imdb-wiki/
This site contains 500k+ 's facial gender and age training data, along with documentation and code, for gender-age projections, adequate resources
Depth estimation, image segmentation:
1, RGBD Training data:
RGBD Training Data list: http://www0.cs.ucl.ac.uk/staff/M.Firman/RGBDdatasets/
Information on deep Learning (1)