{"id":24113,"date":"2021-05-01T05:28:09","date_gmt":"2021-05-01T05:28:09","guid":{"rendered":"https:\/\/rbsitsoftwaresolution.com\/blog\/computer-vision-inches-toward-common-sense-with-facebooks-latest-research.html"},"modified":"2021-05-01T05:28:09","modified_gmt":"2021-05-01T05:28:09","slug":"computer-vision-inches-toward-common-sense-with-facebooks-latest-research","status":"publish","type":"post","link":"https:\/\/www.rbsitsoftwaresolution.in\/blog\/computer-vision-inches-toward-common-sense-with-facebooks-latest-research.html","title":{"rendered":"Computer vision inches toward \u2018common sense\u2019 with Facebook\u2019s latest research"},"content":{"rendered":"

Machine learning is capable of doing all sorts of things as long as you have the data to teach it how. That\u2019s not always easy, and researchers are always looking for a way to add a bit of \u201ccommon sense\u201d to AI so you don\u2019t have to show it 500 pictures of a cat before it gets it. Facebook\u2019s newest research<\/a> takes a big step toward reducing the data bottleneck.<\/p>\n

The company\u2019s formidable AI research division has been working for years now on how to advance and scale things like advanced computer vision algorithms, and has made steady progress, generally shared with the rest of the research community. One interesting development Facebook has pursued in particular is what\u2019s called \u201csemi-supervised learning.\u201d<\/p>\n

Generally when you think of training an AI, you think of something like the aforementioned 500 pictures of cats \u2014 images that have been selected and labeled (which can mean outlining the cat, putting a box around the cat or just saying there\u2019s a cat in there somewhere) so that the machine learning system can put together an algorithm to automate the process of cat recognition. Naturally if you want to do dogs or horses, you need 500 dog pictures, 500 horse pictures, etc. \u2014 it scales linearly<\/em>, which is a word you never want to see in tech.<\/p>\n

Semi-supervised learning, related to \u201cunsupervised\u201d learning, involves figuring out important parts of a data set without any labeled data at all. It doesn\u2019t just go wild, there\u2019s still structure; for instance, imagine you give the system a thousand sentences to study, then showed it 10 more that have several of the words missing. The system could probably do a decent job filling in the blanks just based on what it\u2019s seen in the previous thousand. But that\u2019s not so easy to do with images and video \u2014 they aren\u2019t as straightforward or predictable.<\/p>\n

But Facebook researchers have shown that while it may not be easy, it\u2019s possible and in fact very effective. The DINO system (which stands rather unconvincingly for \u201cDIstillation of knowledge with NO labels\u201d) is capable of learning to find objects of interest in videos of people, animals and objects quite well without any labeled data whatsoever.<\/p>\n

<\/a><\/p>\n

Image Credits:<\/strong> Facebook<\/p>\n

It does this by considering the video not as a sequence of images to be analyzed one by one in order, but as a complex, interrelated set, like the difference between \u201ca series of words\u201d and \u201ca sentence.\u201d By attending to the middle and the end of the video as well as the beginning, the agent can get a sense of things like \u201can object with this general shape goes from left to right.\u201d That information feeds into other knowledge, like when an object on the right overlaps with the first one, the system knows they\u2019re not the same thing, just touching in those frames. And that knowledge in turn can be applied to other situations. In other words, it develops a basic sense of visual meaning, and does so with remarkably little training on new objects.<\/p>\n

This results in a computer vision system that\u2019s not only effective \u2014 it performs well compared with traditionally trained systems \u2014 but more relatable and explainable. For instance, while an AI that has been trained with 500 dog pictures and 500 cat pictures will recognize both, it won\u2019t really have any idea that they\u2019re similar in any way. But DINO \u2014 although it couldn\u2019t be specific \u2014 gets that they\u2019re similar visually to one another, more so anyway than they are to cars, and that metadata and context is visible in its memory. Dogs and cats are \u201ccloser\u201d in its sort of digital cognitive space than dogs and mountains. You can see those concepts as little blobs here \u2014 see how those of a type stick together:<\/p>\n

<\/a><\/p>\n

Image Credits:<\/strong> Facebook<\/p>\n

This has its own benefits, of a technical sort we won\u2019t get into here. If you\u2019re curious, there\u2019s more detail in the papers linked in Facebook\u2019s blog post.<\/p>\n

There\u2019s also an adjacent research project, a training method called PAWS, which further reduces the need for labeled data. PAWS combines some of the ideas of semi-supervised learning with the more traditional supervised method, essentially giving the training a boost by letting it learn from both the labeled and unlabeled data.<\/p>\n

Facebook of course needs good and fast image analysis for its many user-facing (and secret) image-related products, but these general advances to the computer vision world will no doubt be welcomed by the developer community for other purposes.<\/p>\n

\n

WTF is computer vision?<\/a><\/p>\n<\/blockquote>\n

\n

\n

Enhanced computer vision, sensors raise manufacturing stakes for robots as a service<\/a><\/p>\n<\/blockquote>\n

\n

\u00a0<\/p>\n

Article source: http:\/\/feedproxy.google.com\/~r\/techcrunch\/social\/~3\/ORCQZmQD2VM\/<\/p>","protected":false},"excerpt":{"rendered":"

Machine learning is capable of doing all sorts of things as long as you have the data to teach it how. That\u2019s not always easy, and researchers are always looking <\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"jetpack_post_was_ever_published":false,"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[3],"tags":[10,7,14,15,6,12,11],"class_list":["post-24113","post","type-post","status-publish","format-standard","hentry","category-latest-update","tag-computer-news","tag-downloads","tag-microsoft-download-news","tag-networking-updates","tag-technology-update","tag-wordpress-plugin-download","tag-wordpress-theme-download"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t