Skip to content

Commit

Permalink
edits proj-soils
Browse files Browse the repository at this point in the history
  • Loading branch information
@clmarmy
clmarmy committed Apr 25, 2024
1 parent 3380eee commit ee79e57
Show file tree
Hide file tree
Showing 4 changed files with 22 additions and 22 deletions.
6 changes: 3 additions & 3 deletions PROJ-SOILS/index.html
View file
Original file line number Diff line number Diff line change
Expand Up @@ -629,7 +629,7 @@ <h3 id="43-office-federal-de-la-statistique-ofs">4.3 Office Fédéral de la Stat
<p>OFS has also created a deep learning model prototype to automatically segment land cover types. However, different than the models of IGN and HEIG-VD, it works with two steps:</p>
<ol>
<li>The input imagery is processed by the <em>Segment Anything Model (SAM)</em> by Meta <sup id="fnref:9"><a class="footnote-ref" href="#fn:9">9</a></sup>. This step is used to create segments of pixels that belong to the same class without labeling the produced segments.</li>
<li>In a second step, the formerly produced segments are classified using a part of the <em>ADELE</em> pipeline, which is designed to automatize the acquisition of the <a href="https://www.bfs.admin.ch/bfs/fr/home/statistiques/espace-environnement/enquetes/area.html"><em>Statistique suisse de la superficie</em></a>, a point sampling grid with a mesh width of 100 meters. The data points on this grid are only representative for the exact coordinate that they lay on, not for the 100 meter square they are in, which means that the ground truth cannot be treated as an image <sup id="fnref2:10"><a class="footnote-ref" href="#fn:10">10</a></sup>. As a result, their model has been trained to classify the land cover class of only the centre pixel in an input image. To classify a segment, then, the model is used to classify a small number of sample pixels of a segment and the most frequent class is chosen as the class of the segment. The used architecture is a ConvNeXtLarge<sup id="fnref:11"><a class="footnote-ref" href="#fn:11">11</a></sup> architecture, which is a CNN. The model is limited to RGB imagery only and has been trained with a spatial resolution of 25 cm.</li>
<li>In a second step, the formerly produced segments are classified using a part of the <em>ADELE</em> pipeline, which is designed to automatize the acquisition of the <a href="https://www.bfs.admin.ch/bfs/fr/home/statistiques/espace-environnement/enquetes/area.html"><em>Statistique suisse de la superficie</em></a>, a point sampling grid with a mesh width of 100 meters. The data points on this grid are only representative for the exact coordinate that they lay on, not for the 100 meter square they are in, which means that the ground truth cannot be treated as an image <sup id="fnref:10"><a class="footnote-ref" href="#fn:10">10</a></sup>. As a result, their model has been trained to classify the land cover class of only the centre pixel in an input image. To classify a segment, then, the model is used to classify a small number of sample pixels of a segment and the most frequent class is chosen as the class of the segment. The used architecture is a ConvNeXtTiny<sup id="fnref:11"><a class="footnote-ref" href="#fn:11">11</a></sup> architecture, which is a CNN. The model is limited to RGB imagery only and has been trained with a spatial resolution of 25 cm.</li>
</ol>
<h2 id="5-methodology">5. Methodology<a class="headerlink" href="#5-methodology" title="Permanent link">&para;</a></h2>
<p>The Methodology section describes the infrastructure used to run the models and to reproduce the project. Furthermore, it describes precisely the evaluation and fine-tuning approaches.</p>
Expand Down Expand Up @@ -1042,7 +1042,7 @@ <h3 id="71-evaluation">7.1 Evaluation<a class="headerlink" href="#71-evaluation"
<p><strong>HEIG-VD</strong>
The <em>HEIG-VD</em> model, although it is outperformed by the other two institutions' models on Extent 1, performs significantly better in masked Extent 1 and in Extent 2. The model also performed best in the qualitative assessment. The assessment of the performance in Extent 2 shows that the square artefacts are responsible for a great share of false predictions.</p>
<p><strong>OFS</strong>
The OFS model <em>OFS_ADELE2(+SAM)</em> performs similarly to the best-performing IGN model, its outputs are not prone to square artefacts, and the inferences are very clean due to its usage of the SAM model. The downside of the OFS model is that it is specifically adapted for the <em>Statistique suisse de la superficie</em><sup id="fnref:10"><a class="footnote-ref" href="#fn:10">10</a></sup> and thus cannot be retrained on a different dataset.</p>
The OFS model <em>OFS_ADELE2(+SAM)</em> performs similarly to the best-performing IGN model, its outputs are not prone to square artefacts, and the inferences are very clean due to its usage of the SAM model.</p>
<p>The goal of the evaluation phase was to identify the most promising model for further steps in the project. Based on the results of the evaluation, the HEIG-VD model was chosen. It performed best in masked Extent 1 and in Extent 2, and it performed best in the qualitative assessment. Additionally, the model needs only aerial imagery with the three RGB channels which allows for an easier reproducibility. The model weights and source code of the HEIG-VD model were kindly shared with us, which enabled us to fine-tune the model to adapt to the specifics of this project. However, the premise of choosing the HEIG-VD model was that we are able to mitigate the square artefacts to an acceptable degree.</p>
<h3 id="72-fine-tuning">7.2 Fine-Tuning<a class="headerlink" href="#72-fine-tuning" title="Permanent link">&para;</a></h3>
<p>The following keypoints can be extracted from the fine-tuning results:</p>
Expand Down Expand Up @@ -1203,7 +1203,7 @@ <h2 id="10-bibliography">10. Bibliography<a class="headerlink" href="#10-bibliog
<p>Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, Piotr Dollár, and Ross Girshick. Segment Anything. April 2023. arXiv:2304.02643 [cs]. URL: <a href="http://arxiv.org/abs/2304.02643">http://arxiv.org/abs/2304.02643</a> (visited on 2024-04-05), <a href="https://doi.org/10.48550/arXiv.2304.02643">doi:10.48550/arXiv.2304.02643</a>.&#160;<a class="footnote-backref" href="#fnref:9" title="Jump back to footnote 9 in the text">&#8617;</a></p>
</li>
<li id="fn:10">
<p>Unknown. <em>Arealstatistik Schweiz. Erhebung der Bodennutzung und der Bodenbedeckung. (Ausgabe 2019 / 2020)</em>. Number 9406112. Bundesamt für Statistik (BFS), Neuchâtel, September 2019. Backup Publisher: Bundesamt für Statistik (BFS). URL: <a href="https://dam-api.bfs.admin.ch/hub/api/dam/assets/9406112/master">https://dam-api.bfs.admin.ch/hub/api/dam/assets/9406112/master</a>.&#160;<a class="footnote-backref" href="#fnref:10" title="Jump back to footnote 10 in the text">&#8617;</a><a class="footnote-backref" href="#fnref2:10" title="Jump back to footnote 10 in the text">&#8617;</a></p>
<p>Unknown. <em>Arealstatistik Schweiz. Erhebung der Bodennutzung und der Bodenbedeckung. (Ausgabe 2019 / 2020)</em>. Number 9406112. Bundesamt für Statistik (BFS), Neuchâtel, September 2019. Backup Publisher: Bundesamt für Statistik (BFS). URL: <a href="https://dam-api.bfs.admin.ch/hub/api/dam/assets/9406112/master">https://dam-api.bfs.admin.ch/hub/api/dam/assets/9406112/master</a>.&#160;<a class="footnote-backref" href="#fnref:10" title="Jump back to footnote 10 in the text">&#8617;</a></p>
</li>
<li id="fn:11">
<p>Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, and Saining Xie. A ConvNet for the 2020s. March 2022. arXiv:2201.03545 [cs]. URL: <a href="http://arxiv.org/abs/2201.03545">http://arxiv.org/abs/2201.03545</a> (visited on 2024-03-21), <a href="https://doi.org/10.48550/arXiv.2201.03545">doi:10.48550/arXiv.2201.03545</a>.&#160;<a class="footnote-backref" href="#fnref:11" title="Jump back to footnote 11 in the text">&#8617;</a></p>
Expand Down
2 changes: 1 addition & 1 deletion search/search_index.json
View file

Large diffs are not rendered by default.

36 changes: 18 additions & 18 deletions sitemap.xml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,92 +2,92 @@
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
<url>
<loc>None</loc>
<lastmod>2024-04-24</lastmod>
<lastmod>2024-04-25</lastmod>
<changefreq>daily</changefreq>
</url>
</urlset>
Binary file modified sitemap.xml.gz
View file
Binary file not shown.

0 comments on commit ee79e57

Please sign in to comment.