https://blog.defectiv.es/en/tags/deep-learning/Recent content in deep-learning on Gaia Lab · BlogHugoen-GBWed, 23 Sep 2026 09:00:00 +0200https://blog.defectiv.es/en/posts/segmentacion-dental-y-100-semillas/Tue, 22 Sep 2026 13:30:00 +0200https://blog.defectiv.es/en/posts/segmentacion-dental-y-100-semillas/<p>Sixth instalment of <strong>Cluster X-ray</strong>. After the <a href="https://blog.defectiv.es/en/posts/arboles-uno-a-uno-desde-nubes-de-puntos-lidar/">forests</a> , the most frugal line in the archive: twenty GPU-hours in four months and a methodological question that many papers publish without answering.</p> <h2 id="the-question">The question <a class="hanchor" href="#the-question" aria-label="Enlace a esta sección">#</a></h2> <p>Automatically segmenting structures in dental images, separating the region of interest from the background pixel by pixel, is the basis of any subsequent measurement or assisted diagnosis. The literature compares architectures with a single run per model, and the differences between them are often on the order of one point. The line asks two things: <strong>which architecture segments this type of image best, and how much of the difference between two models is real and how much is the lottery of initialisation.</strong></p>https://blog.defectiv.es/en/posts/arboles-uno-a-uno-desde-nubes-de-puntos-lidar/Tue, 22 Sep 2026 13:15:00 +0200https://blog.defectiv.es/en/posts/arboles-uno-a-uno-desde-nubes-de-puntos-lidar/<p>Fifth instalment of <strong>Cluster X-ray</strong>. After the <a href="https://blog.defectiv.es/en/posts/4400-complejos-de-proteinas-con-boltz-2/">proteins</a> , another line unrelated to language: forests.</p> <h2 id="the-question">The question <a class="hanchor" href="#the-question" aria-label="Enlace a esta sección">#</a></h2> <p>A LiDAR scan of a forest plot returns millions of unlabelled points. To estimate above-ground biomass, and with it the stored carbon, you need to know which points belong to which tree: splitting the forest into <strong>individuals</strong>. It is a 3D instance segmentation problem, hard wherever the crowns touch. The line asks <strong>which architecture does it best, how much the result depends on the luck of the initialisation, and which information in the cloud really matters.</strong></p>