{"slug": "anaximander-interactively-running-geospatial-deep-learning-models-on-any-compute", "title": "Anaximander: Interactively Running Geospatial Deep Learning Models on Any Compute Backend", "summary": "Microsoft researchers released Anaximander, an open-source system that runs geospatial deep learning models on any compute backend behind one interactive interface, with code at github.com/microsoft/nxmndr. The system pairs an inference server — providing session management, model caching, and per-tile result streaming — with a QGIS plugin that handles tiling, result reassembly, georeferencing, and real-time per-tile status visualization, plus a user-interface path that injects layer legends as prompts into vision-language models. The team demonstrated a code-free side-by-side comparison of three heterogeneous models on an agricultural field delineation task: gpt-image-1 via a cloud API, Segment Anything Model 3 (SAM3) on a remote GPU, and DelineateAnything on a local CPU.", "body_md": "arXiv:2610.09085v1 Announce Type: new \nAbstract: Applying deep learning models to satellite imagery from within geographic information systems (GIS) remains high-friction for remote sensing practitioners. Models arrive in incompatible formats and target different compute environments, from local workstations to serverless cloud services. As a result, every evaluation demands custom deployment, tiling, and georeferencing code before a single prediction reaches the analyst's map. This friction discourages systematic comparison in a domain where model choice directly affects operational outcomes such as field delineation, crop monitoring, and disaster response. We present Anaximander, an open-source system that unifies model source and compute location choice behind one interactive interface. The system's backend is an inference server that loads models from multiple commonly-used sources and serves them on any accessible compute backend. The server provides session management and model caching, and streams results back per tile. The backend is paired with a QGIS plugin that drives tiling, result reassembly, georeferencing, and real-time per-tile status visualization. An additional user-interface path injects layer legends as prompts into vision-language models. We demonstrate the system in a code-free side-by-side comparison of three heterogeneous models on an agricultural field delineation task: gpt-image-1 via a cloud API, Segment Anything Model 3 (SAM3) on a remote GPU, and DelineateAnything on a local CPU. The inference backend and protocol are open-source and available at https://github.com/microsoft/nxmndr.", "url": "https://wpnews.pro/news/anaximander-interactively-running-geospatial-deep-learning-models-on-any-compute", "canonical_source": "https://arxiv.org/abs/2610.09085", "published_at": "2026-10-08 04:00:00+00:00", "updated_at": "2026-10-08 04:20:03.364793+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "computer-vision", "ai-tools", "developer-tools"], "entities": ["Microsoft", "Anaximander", "QGIS", "gpt-image-1", "Segment Anything Model 3", "SAM3", "DelineateAnything", "github.com/microsoft/nxmndr"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/anaximander-interactively-running-geospatial-deep-learning-models-on-any-compute", "markdown": "https://wpnews.pro/news/anaximander-interactively-running-geospatial-deep-learning-models-on-any-compute.md", "text": "https://wpnews.pro/news/anaximander-interactively-running-geospatial-deep-learning-models-on-any-compute.txt", "jsonld": "https://wpnews.pro/news/anaximander-interactively-running-geospatial-deep-learning-models-on-any-compute.jsonld"}}