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="/about">About</a></li> 19<li><a href="https://roundabout-host.com/roundabout">Roundabout-host</a></li> 20</ul> 21<ul> 22<li><a href="mailto:root@roundabout-host.com" id="mail-link">root@roundabout-host.com</a></li> 23</ul> 24</nav> 25</header> 26<main> 27 28<div class="project-title"> 29<h1> 30Waste detection 31</h1> 32<a href="https://roundabout-host.com/roundabout/waste-detection" class="repository-button">Go to repository</a> 33</div> 34<p class="tags"> 35 36<a href="/index/ai.html" class="tag">ai</a> 37 38<a href="/index/machine learning.html" class="tag">machine learning</a> 39 40<a href="/index/computer vision.html" class="tag">computer vision</a> 41 42<a href="/index/environment.html" class="tag">environment</a> 43 44<a href="/index/waste.html" class="tag">waste</a> 45 46<a href="/index/data.html" class="tag">data</a> 47 48<a href="/index/yolo.html" class="tag">yolo</a> 49 50<a href="/index/keras.html" class="tag">keras</a> 51 52<a href="/index/python.html" class="tag">python</a> 53 54<a href="/index/agpl.html" class="tag">agpl</a> 55 56<a href="/index/cc-by-sa.html" class="tag">cc-by-sa</a> 57 58<a href="/index/cc-by-nc.html" class="tag">cc-by-nc</a> 59 60<a href="/index/waste detection.html" class="tag">waste detection</a> 61 62<a href="/index/creative commons.html" class="tag">creative commons</a> 63 64<a href="/index/crowdsourcing.html" class="tag">crowdsourcing</a> 65 66</p> 67<article class="content-area"> 68<p>An YOLO model specifically trained to detect waste in images. We collected a custom dataset of waste, 69classified and annotated it, and trained the model using YOLOv8 on Keras. The model is able to detect 70the position and class of waste in images. It can separate the following classes of waste: 71</p><ul><li><p>Plastic 72</p></li><li><p>Metal 73</p></li><li><p>Paper 74</p></li><li><p>Glass 75</p></li><li><p>Organic 76</p></li><li><p>Wood 77</p></li><li><p>Textile 78</p></li><li><p>Medical 79</p></li><li><p>Electronic 80</p></li><li><p>Other 81</p></li></ul><p>There are actually over 90 classes, but they are grouped into these ones because there is only 82a small amount (in the 1-2 digits) of samples for each class. The model is able to detect multiple 83classes in the same image, or acknowledge that there is no waste in the image. 84</p><p>YOLO is used pretrained on the COCO dataset, and then fine-tuned on our dataset. This is until 85there are enough samples for each class to train the model from scratch. 86</p><p>Contributions are welcome. If you have a dataset of waste, please share it with us. We will 87annotate it (if it's not already) and train the model on it. 88</p><p>You don't need technical knowledge to contribute. If you can take photos with a phone and make 89a ZIP of them, you can contribute. If you can draw rectangles around objects with the mouse, 90you can contribute. If you can drop photos into folders corresponding to their class, you can 91contribute. 92</p><p>Of course, if you do have technical knowledge, you can help with the model training, testing, and 93optimisation. You can also improve the training code and better process the dataset. 94</p><h2>Licence</h2><p>Unless otherwise stated, the software (source code) in this repository is licenced under the GNU Affero General Public License, 95with an additional permission in section 7 - basically an LAGPL. 96A copy of the licence is located in the file <code>LICENCE-GNU-AGPL.md</code> in this directory. 97</p><blockquote><p>This program is free software, you can redistribute it and/or modify 98it under the terms of GNU Affero General Public License 99as published by the Free Software Foundation, either version 3 100of the License, or (at your option) any later version. 101</p><p>This program is distributed in the hope that it will be useful, 102but WITHOUT ANY WARRANTY; without even the implied warranty of 103or FITNESS FOR A PARTICULAR PURPOSE. See the 104General Public License for more details. 105</p><p>You should have received a copy of the the GNU Affero 106General Public License, along with this program. If not, see 107<a href="https://www.gnu.org/licenses/">https://www.gnu.org/licenses/</a>. 108</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 109combining it with other code, such other code is not for that reason 110alone subject to any of the requirements of the GNU Affero GPL 111version 3. 112</p></blockquote><p>Unless otherwise stated, the annotations in this repository are dual-licensed under either of these licences: 113</p><ul><li><p><a href="https://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution Share-Alike 4.0</a>. 114To 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/ 115or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. 116</p></li><li><p><a href="https://creativecommons.org/licenses/by-nc/4.0/">Creative Commons Attribution Non-Commercial 4.0</a>. 117To 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/ 118or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA. 119</p></li></ul><p>You do not have to comply with the terms of both licences. Choose one and comply only with it. 120</p><p>The images in the repository are licensed depending on their source: 121</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). 122</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>. 123</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>. 124</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. 125</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. 126</p></li></ul><p>I hope I didn't infringe on any licences or miss any sources. If you think there is a problem, 127please send me an email at <a href="mailto:root&#64;roundabout-host.com">root&#64;roundabout-host.com</a> and I'll remove the infringing files. 128</p><p>The VIA application shipped in this repository (located at the <code>/via</code> path) is licensed under the BSD licence. 129</p><blockquote><p>Copyright (c) 2016-2021, Abhishek Dutta, Visual Geometry Group, Oxford University and VIA Contributors. 130All rights reserved. 131Redistribution and use in source and binary forms, with or without 132modification, are permitted provided that the following conditions are met: 133</p><p>Redistributions of source code must retain the above copyright notice, this 134list of conditions and the following disclaimer. 135</p><p>Redistributions in binary form must reproduce the above copyright notice, 136this list of conditions and the following disclaimer in the documentation 137and/or other materials provided with the distribution. 138</p><p>THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 139AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 140IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 141ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE 142LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 143CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 144SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 145INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 146CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 147ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 148POSSIBILITY OF SUCH DAMAGE. 149</p></blockquote> 150</article> 151 152</main> 153<footer> 154<p>Page generated on Sunday, 2 February 2025 at 21:30:11</p> 155<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> 156<p>Hosted at <a href="https://roundabout-host.com/roundabout">Roundabout-host</a> using the static site service, and generated with <a href="/projects/ampoule.html">Ampoule</a>.</p> 157<a href="#">Back to top</a> 158</footer> 159</body> 160</html>