Recurrent Neural Network Writes Sentences About Images | Two Minute Papers #23

This strategy is a mix of two impressive machine learning algorithms:
– convolutional neural networks are outstanding at image classification, i.e., getting out what is noticed on an input image,
– recurrent neural networks that are able of processing a sequence of inputs and outputs, therefore it can produce sentences of what is noticed on the image.

Combining these two strategies tends to make it attainable for a computer to explain in a sentence what is noticed on an input image.

_____________________

The paper “Deep Visible-Semantic Alignments for Producing Image Descriptions” is offered right here:
http://cs.stanford.edu/people today/karpathy/deepimagesent/

A gallery with more effects with the same algorithm:
http://cs.stanford.edu/people today/karpathy/deepimagesent/generationdemo/

You can practice your have convolutional neural network right here:
http://cs.stanford.edu/people today/karpathy/convnetjs/demo/cifar10.html

The supply code for the job is now offered right here:
https://github.com/karpathy/neuraltalk2

Subscribe if you would like to see more of these! – http://www.youtube.com/membership_heart?increase_user=keeroyz

The thumbnail image qualifications was produced by Georgie Pauwels (CC BY two.) – https://flic.kr/p/qrRciQ
Splash display screen/thumbnail design and style: Felícia Fehér – http://felicia.hu

Károly Zsolnai-Fehér’s back links:
Patreon → https://www.patreon.com/TwoMinutePapers
Fb → https://www.fb.com/TwoMinutePapers/
Twitter → https://twitter.com/karoly_zsolnai
World-wide-web → https://cg.tuwien.ac.at/~zsolnai/