﻿{"id":2434,"date":"2021-02-25T14:28:32","date_gmt":"2021-02-25T14:28:32","guid":{"rendered":"https:\/\/uni.hi.is\/helmut\/?p=2434"},"modified":"2021-02-27T14:50:30","modified_gmt":"2021-02-27T14:50:30","slug":"object-detection-using-neural-networks-in-your-smartphone-trained-by-a-supercomputer","status":"publish","type":"post","link":"https:\/\/uni.hi.is\/helmut\/2021\/02\/25\/object-detection-using-neural-networks-in-your-smartphone-trained-by-a-supercomputer\/","title":{"rendered":"Stafr\u00e6ni H\u00e1sk\u00f3ladagurinn 2021: Object detection using neural networks in your smartphone trained by a supercomputer"},"content":{"rendered":"<p>The University of Iceland's Computer Science department is researching <a href=\"https:\/\/uni.hi.is\/helmut\/research\/horizon-2020-future-and-emerging-technologies-programme-deep-extreme-scale-technologies-deep-est\/\">machine learning using the next generation's supercomputer DEEP-EST<\/a> -- by the way: we are also offering a <a href=\"https:\/\/ugla.hi.is\/kennsluskra\/index.php?tab=nam&amp;chapter=namsleid&amp;id=080713_20216&amp;kennsluar=2021&amp;lina=9164\">Data Science specialisation<\/a> in our <a href=\"https:\/\/www.hi.is\/tolvunarfraedi\">Computer Science programme<\/a>, where, e.g., machine learning including deep neural networks is covered. To showcase what is possible if you have a supercomputer to train neural networks, we offer a web page that allows you to use the camera of your smartphone (or laptop) to detect objects in real-time.<\/p>\n<p><a href=\"https:\/\/nvndr.csb.app\/\"><img loading=\"lazy\" decoding=\"async\" class=\"alignright size-thumbnail wp-image-1728\" src=\"http:\/\/uni.hi.is\/helmut\/files\/2019\/10\/qrcode_codesandbox-150x150.png\" alt=\"https:\/\/nvndr.csb.app\/\" width=\"150\" height=\"150\" srcset=\"https:\/\/uni.hi.is\/helmut\/files\/2019\/10\/qrcode_codesandbox-150x150.png 150w, https:\/\/uni.hi.is\/helmut\/files\/2019\/10\/qrcode_codesandbox.png 404w\" sizes=\"auto, (max-width: 150px) 100vw, 150px\" \/><\/a><\/p>\n<p>Just open the following web page and allow your browser to use the camera: <a href=\"https:\/\/nvndr.csb.app\/\">https:\/\/nvndr.csb.app\/<\/a><br \/>\n(Allow up to approx. 1 minute for loading the trained neural network and for initialisation. Web page works best in landscape orientation.)<\/p>\n<p>While neural networks are still best trained on a supercomputer, such as <a href=\"https:\/\/www.deep-projects.eu\">DEEP-EST<\/a> with its Data Analysis Module, the trained neural network even runs in the browser of a smartphone (purely running locally as Javascript in your browser without any connection to a supercomputer, i.e. completely offline after having downloaded the Javascript code and the trained neural network).\n<\/p>\n<p>The used approach is <a href=\"https:\/\/medium.com\/@ManishChablani\/ssd-single-shot-multibox-detector-explained-38533c27f75f\">Single Shot Detector (SSD)<\/a> (the percentage shows how sure the neural network is about the classification) using the <a href=\"https:\/\/towardsdatascience.com\/review-mobilenetv1-depthwise-separable-convolution-light-weight-model-a382df364b69\">MobileNet neural network<\/a> architecture. The dataset used for training is COCO (Common Objects in Context), i.e. only objects of the labeled <a href=\"http:\/\/cocodataset.org\/#explore\">object classes contained in COCO<\/a> will get detected. The <a href=\"https:\/\/hackernoon.com\/tensorflow-js-real-time-object-detection-in-10-lines-of-code-baf15dfb95b2\">Javascript code that is running in your browser<\/a> uses <a href=\"https:\/\/www.tensorflow.org\/lite\/models\/object_detection\/overview\">Tensorflow Lite<\/a> and its <a href=\"https:\/\/github.com\/tensorflow\/models\/blob\/master\/research\/object_detection\/README.md\">Object Detection API<\/a>.<\/p>\n<div id=\"attachment_1977\" style=\"width: 680px\" class=\"wp-caption alignleft\"><img loading=\"lazy\" decoding=\"async\" aria-describedby=\"caption-attachment-1977\" src=\"http:\/\/uni.hi.is\/helmut\/files\/2020\/02\/EQaDGgnXkAEr00B-670x503.jpeg\" alt=\"\" width=\"670\" height=\"503\" class=\"size-medium wp-image-1977\" srcset=\"https:\/\/uni.hi.is\/helmut\/files\/2020\/02\/EQaDGgnXkAEr00B-670x503.jpeg 670w, https:\/\/uni.hi.is\/helmut\/files\/2020\/02\/EQaDGgnXkAEr00B-1024x768.jpeg 1024w, https:\/\/uni.hi.is\/helmut\/files\/2020\/02\/EQaDGgnXkAEr00B-150x113.jpeg 150w, https:\/\/uni.hi.is\/helmut\/files\/2020\/02\/EQaDGgnXkAEr00B-768x576.jpeg 768w, https:\/\/uni.hi.is\/helmut\/files\/2020\/02\/EQaDGgnXkAEr00B-1536x1152.jpeg 1536w, https:\/\/uni.hi.is\/helmut\/files\/2020\/02\/EQaDGgnXkAEr00B-2048x1536.jpeg 2048w\" sizes=\"auto, (max-width: 670px) 100vw, 670px\" \/><p id=\"caption-attachment-1977\" class=\"wp-caption-text\">Example object detection via a neural network<\/p><\/div>\n<p>If you want learn more about the DEEP-EST project where the next generation supercomputer is developed, have a look at the poster below (click on the picture below for PDF version):<\/p>\n<p><a href=\"http:\/\/uni.hi.is\/helmut\/files\/2019\/10\/DEEPEST_poster_A0.pdf\"><img loading=\"lazy\" decoding=\"async\" src=\"http:\/\/uni.hi.is\/helmut\/files\/2019\/10\/DEEPEST_poster.png\" alt=\"PDF of DEEP-EST poster\" class=\"alignright size-full wp-image-1738\" width=\"662\" height=\"933\" srcset=\"https:\/\/uni.hi.is\/helmut\/files\/2019\/10\/DEEPEST_poster.png 662w, https:\/\/uni.hi.is\/helmut\/files\/2019\/10\/DEEPEST_poster-106x150.png 106w, https:\/\/uni.hi.is\/helmut\/files\/2019\/10\/DEEPEST_poster-475x670.png 475w\" sizes=\"auto, (max-width: 662px) 100vw, 662px\" \/> <\/p>\n","protected":false},"excerpt":{"rendered":"<p>The University of Iceland's Computer Science department is researching machine learning using the next generation's supercomputer DEEP-EST -- by the way: we are also offering a Data Science specialisation in our Computer Science programme, where, e.g., machine learning including deep neural networks is covered. To showcase what is possible if you have a supercomputer to [&hellip;]<\/p>\n","protected":false},"author":512,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[79],"tags":[],"class_list":["post-2434","post","type-post","status-publish","format-standard","hentry","category-research"],"_links":{"self":[{"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/posts\/2434","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/users\/512"}],"replies":[{"embeddable":true,"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/comments?post=2434"}],"version-history":[{"count":21,"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/posts\/2434\/revisions"}],"predecessor-version":[{"id":2464,"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/posts\/2434\/revisions\/2464"}],"wp:attachment":[{"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/media?parent=2434"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/categories?post=2434"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/uni.hi.is\/helmut\/wp-json\/wp\/v2\/tags?post=2434"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}