Some resources on understanding Images with image processing, computer vision and machine learning.
- JavaScript (p5.js) image processing examples
- ofxCv, an openFrameworks addon for computer vision tasks
Excellent, compact explanation of Convolutional Neural Networks:
DCGAN Generative Cat Faces (Link)(Link)
- DCGAN Original repo https://github.com/Newmu/dcgan_code
- DCGAN Manga characters http://mattya.github.io/chainer-DCGAN/
.@richardsocher at MetaMind is adding reasoning to AI. Give it a photo, ask it questions, it will tell what’s what. pic.twitter.com/ihWUH6w7YH
— Kevin Kelly (@kevin2kelly) March 10, 2016
t-SNE Grids:
w/ ofxCcv ofxTSNE ofxAssignment your basically halfway there to a dope thesis project….
— zach lieberman (@zachlieberman) January 28, 2016
T-Sne converging:
t-SNE point cloud converging onto 2d https://t.co/pcyKzW38GW pic.twitter.com/5Eeu9DGlVD
— Gene Kogan (@genekogan) January 31, 2016
Hungarian grid layout algorithm:
t-SNE point cloud converging onto 2d https://t.co/pcyKzW38GW pic.twitter.com/5Eeu9DGlVD
the hungarian algorithm is a solution to the assignment problem. turn clouds into grids: https://t.co/NXakp3peuG pic.twitter.com/z2mqNajKXl
— Kyle McDonald (@kcimc) January 23, 2016
- Gene Kogan, Flowers
- Quasimondo, Sets1, Sets2
- Tom White, neural font experiment (based on Erik Bernhardsson’s work)
Gulcan Can, et al., Mayan Glyphs:
mayan glyphs clustered with t-SNE https://t.co/oedelRitoR pic.twitter.com/rLKBl6evTZ
— Kyle McDonald (@kcimc) February 24, 2016
Style Transfer:
Examples (From BoredPanda):
Colorizing B&W movies:
Colorizing B/W Movies with Neural Nets: https://t.co/4Ly7NmBGKZ
Testing Ryan Dahl’s Net on C.Chaplin’s “The Kid” pic.twitter.com/2f2Nd1GzP7— samim (@samim) January 8, 2016
Colorizing Black and White Photos with deep learning: https://t.co/AwK8mfjHrs pic.twitter.com/SBvufSX8zS
— samim (@samim) January 8, 2016
hi-res stylenet w/ @mtyka’s tiling code https://t.co/plB4vv59Ox = https://t.co/ub1Cl9osuy + https://t.co/XvtMkOSzHJ pic.twitter.com/OGM0KdryXe
— Gene Kogan (@genekogan) January 18, 2016
Tools
Temboo API’s for obtaining images:
- https://temboo.com/library/Library/Flickr/
- https://temboo.com/library/Library/eBay/Finding/FindItemsByImage/
- https://temboo.com/library/Library/Google/Picasa/
- …plus Instagram, Twitter, Facebook, YouTube, NYTimes…
Some machine learning tools for images:
- Gene Kogan, ofxTsne Gridding Example
- Memo Akten, ofxMSATensorFlow
- Jetpac’s DeepBeliefSDK (in-browser demo)
- Clarifai (Visual Recognition API)
- Kyle McDonald, ofxAssignment
Word2Vec Readings:
- https://quomodocumque.wordpress.com/2016/01/15/messing-around-with-word2vec/
- http://bookworm.benschmidt.org/posts/2015-10-25-Word-Embeddings.html
Yet more datasets
- https://twitter.com/hardmaru/status/691104567367315456
- https://twitter.com/moebio/status/687693885343137792
- https://twitter.com/samim/status/685763173497073664
- https://twitter.com/dantasse/status/684801482839863296
- https://twitter.com/alexczet/status/684784505182367744
- https://twitter.com/mgiraldo/status/684735366046330881
- https://twitter.com/quasimondo/status/684695813549854720
- https://twitter.com/textfiles/status/680830601453301762
- https://twitter.com/samim/status/677371074376527872
Readings
- Gene Kogan, From Pixels to Paragraphs: How artistic experiments with deep learning guard us from hype
- Julia Evans: How to trick a neural network into thinking a panda is a vulture
- Artem Khurshudov, Suddenly, a leopard print sofa appears
- Mike Tyka, The art of neural networks (video)
- Alexander Mordvintsev, Christopher Olah, and Mike Tyka (Google). Inceptionism.
- Tomasz Malisiewicz. From feature descriptors to deep learning: 20 years of computer vision
- Samim Winiger, Generating Stories about Images
- Samim Winiger, Adversarial Machines
- Kyle McDonald, A Return to Machine Learning (draft)
- Memo Akten, Review of machine / deep learning in an artistic context
