waste.html
HTML document, Unicode text, UTF-8 text
1<!DOCTYPE html> 2<html lang="en"> 3<head> 4<meta charset="UTF-8"> 5<title> 6Waste detection 7</title> 8<link rel="stylesheet" href="/static/style.css"> 9<meta name="viewport" content="width=device-width, initial-scale=1.0"> 10</head> 11<body> 12<header> 13<nav> 14<ul> 15<li><a href="/">Home</a></li> 16<li><a href="/projects">Projects</a></li> 17<li><a href="/index">Index</a></li> 18<li><a href="https://roundabout-host.com/roundabout">Roundabout-host</a></li> 19</ul> 20<ul> 21<li><a href="mailto:root@roundabout-host.com" id="mail-link">root@roundabout-host.com</a></li> 22</ul> 23</nav> 24</header> 25<main> 26 27<h1 class="project-title"> 28<span>Waste detection</span> 29<a href="https://roundabout-host.com/roundabout/waste-detection">Repository</a> 30</h1> 31<article class="content-area"> 32<p>An YOLO model specifically trained to detect waste in images. We collected a custom dataset of waste, 33classified and annotated it, and trained the model using YOLOv8 on Keras. The model is able to detect 34the position and class of waste in images. It can separate the following classes of waste: 35</p><ul><li><p>Plastic 36</p></li><li><p>Metal 37</p></li><li><p>Paper 38</p></li><li><p>Glass 39</p></li><li><p>Organic 40</p></li><li><p>Wood 41</p></li><li><p>Textile 42</p></li><li><p>Medical 43</p></li><li><p>Electronic 44</p></li><li><p>Other 45</p></li></ul><p>There are actually over 90 classes, but they are grouped into these ones because there is only 46a small amount (in the 1-2 digits) of samples for each class. The model is able to detect multiple 47classes in the same image, or acknowledge that there is no waste in the image. 48</p><p>YOLO is used pretrained on the COCO dataset, and then fine-tuned on our dataset. This is until 49there are enough samples for each class to train the model from scratch. 50</p><p>Contributions are welcome. If you have a dataset of waste, please share it with us. We will 51annotate it (if it's not already) and train the model on it. 52</p><p>You don't need technical knowledge to contribute. If you can take photos with a phone and make 53a ZIP of them, you can contribute. If you can draw rectangles around objects with the mouse, 54you can contribute. If you can drop photos into folders corresponding to their class, you can 55contribute. 56</p><p>Of course, if you do have technical knowledge, you can help with the model training, testing, and 57optimisation. You can also improve the training code and better process the dataset. 58</p><h2>Licence</h2><p>Unless otherwise stated, the software (source code) in this repository is licenced under the GNU Affero General Public License, 59with an additional permission in section 7 - basically an LAGPL. 60A copy of the licence is located in the file <code>LICENCE-GNU-AGPL.md</code> in this directory. 61</p><blockquote><p>This program is free software, you can redistribute it and/or modify 62it under the terms of GNU Affero General Public License 63as published by the Free Software Foundation, either version 3 64of the License, or (at your option) any later version. 65</p><p>This program is distributed in the hope that it will be useful, 66but WITHOUT ANY WARRANTY; without even the implied warranty of 67or FITNESS FOR A PARTICULAR PURPOSE. See the 68General Public License for more details. 69</p><p>You should have received a copy of the the GNU Affero 70General Public License, along with this program. If not, see 71<https://www.gnu.org/licenses/>. 72</p><h5>Additional permission under the GNU Affero GPL version 3 section 7:</h5><p>If you modify this program, or any covered work, by linking or 73combining it with other code, such other code is not for that reason 74alone subject to any of the requirements of the GNU Affero GPL 75version 3. 76</p></blockquote><p>Unless otherwise stated, the annotations in this repository are dual-licensed under either of these licences: 77</p><ul><li><p><a href="https://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution Share-Alike 4.0</a>. 78To view a copy of this license, read the <code>LICENCE-CC-BY-SA.md</code> file, visit http://creativecommons.org/licenses/by-sa/4.0/ 79or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. 80</p></li><li><p><a href="https://creativecommons.org/licenses/by-nc/4.0/">Creative Commons Attribution Non-Commercial 4.0</a>. 81To view a copy of this license, read the <code>LICENCE-CC-BY-NC.md</code> file, visit http://creativecommons.org/licenses/by-nc/4.0/ 82or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. 83</p></li></ul><p>You do not have to comply with the terms of both licences. Choose one and comply only with it. 84</p><p>The images in the repository are licensed depending on their source: 85</p><ul><li><p>If the file name has the <code>original</code> prefix (or no prefix at all), the images are licenced under the same terms as the annotations (see above). 86</p></li><li><p>If the file name has the <code>trash-detection</code> prefix, it comes from https://universe.roboflow.com/nora-slimani/trash-detection-otdmj and is licensed under <a href="https://creativecommons.org/licenses/by/4.0/">Creative Commons Attribution 4.0</a>. 87</p></li><li><p>If the file name has the <code>ffml</code> prefix, it comes from https://www.kaggle.com/datasets/cristeaioan/ffml-dataset and is licensed under <a href="https://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution Share-Alike 4.0</a>. 88</p></li><li><p>If the file name has the <code>dwc</code> prefix, it comes from https://www.kaggle.com/datasets/arkadiyhacks/drinking-waste-classification and is public domain. 89</p></li><li><p>If the file name has the <code>hitl</code> prefix, it comes from https://www.kaggle.com/datasets/humansintheloop/recycling-dataset and is public domain. 90</p></li></ul><p>I hope I didn't infringe on any licences or miss any sources. If you think there is a problem, 91please send me an email at <mailto:root@roundabout-host.com> and I'll remove the infringing files. 92</p><p>The VIA application shipped in this repository (located at the <code>/via</code> path) is licensed under the BSD licence. 93</p><blockquote><p>Copyright (c) 2016-2021, Abhishek Dutta, Visual Geometry Group, Oxford University and VIA Contributors. 94All rights reserved. 95Redistribution and use in source and binary forms, with or without 96modification, are permitted provided that the following conditions are met: 97</p><p>Redistributions of source code must retain the above copyright notice, this 98list of conditions and the following disclaimer. 99</p><p>Redistributions in binary form must reproduce the above copyright notice, 100this list of conditions and the following disclaimer in the documentation 101and/or other materials provided with the distribution. 102</p><p>THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 103AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 104IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 105ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE 106LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 107CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 108SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 109INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 110CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 111ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 112POSSIBILITY OF SUCH DAMAGE. 113</p></blockquote> 114</article> 115 116</main> 117<footer> 118<p>Page generated on Tuesday, 15 October 2024 at 17:00:35</p> 119<p xmlns:cc="http://creativecommons.org/ns#" >This work is marked with <a href="https://creativecommons.org/publicdomain/zero/1.0/?ref=chooser-v1" target="_blank" rel="license noopener noreferrer" style="display:inline-block;">CC0 1.0 Universal</a> (🄍). No rights reserved.</p> 120<a href="#">Back to top</a> 121</footer> 122</body> 123</html>