Hi Guys, I was wondering if there are deep learning techniques existing which given a photo ( picture ) could analyse and generate a textual description of the photo.
I am referring not just tag generation but rather complete sentence generation about the description of the photo.
http://openaccess.thecvf.com/content_cvpr_2015/papers/Yan_Deep_Correlation_for_2015_CVPR_paper.pdf
https://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Karpathy_Deep_Visual-Semantic_Alignments_2015_CVPR_paper.pdf
https://www.jair.org/index.php/jair/article/download/10985/26134/
http://openaccess.thecvf.com/content_ICCV_2017/papers/Zhang_StackGAN_Text_to_ICCV_2017_paper.pdf
I guess what you mentioned is Image caption, that is, given an image for a network, and the output is a description of the image. Also, this process can be inverse，a network accepts a textual description and generates an related image. Further, an extended field is that Visual Question Answering（VQA）, given the network an image and a question, then it is expected to output the answer.
Hope it helps.
https://news.developer.nvidia.com/ai-can-now-fix-your-grainy-photos-by-only-looking-at-grainy-photos/?ncid=--45511&fbclid=IwAR0uG1_9GM3ddZm_Oyxofy0sssvdeWTV4DjNDpmWcaxRpBqpomdXR7CCI0A
- Badges
- Science topic