RECORD
Anaximander: Interactively Running Geospatial Deep Learning Models on Any Compute Backend
Computer Vision and Pattern Recognition · Submitted 6 Oct 2026
abstract compiled · 2026-10-08 · no language-model review
The authors' main claim We present Anaximander, an open-source system that unifies model source and compute location choice behind one interactive interface.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
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.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Applying deep learning models to satellite imagery from within geographic information systems (GIS) remains high-friction for remote sensing practitioners. | background 0.98 | 0.07 | 0.03 | asserted 1.00 |
| s1 | Models arrive in incompatible formats and target different compute environments, from local workstations to serverless cloud services. | background 0.97 | 0.09 | 0.02 | asserted 1.00 |
| s2 | As a result, every evaluation demands custom deployment, tiling, and georeferencing code before a single prediction reaches the analyst's map. | background 0.89 | 0.08 | 0.03 | asserted 1.00 |
| s3 | This friction discourages systematic comparison in a domain where model choice directly affects operational outcomes such as field delineation, crop monitoring, and disaster response. | background 0.78 | 0.08 | 0.05 | asserted 1.00 |
| s4 | We present Anaximander, an open-source system that unifies model source and compute location choice behind one interactive interface. | contribution 1.00 | 0.91 | 0.35 | asserted 1.00 |
| s5 | The system's backend is an inference server that loads models from multiple commonly-used sources and serves them on any accessible compute backend. | method 0.84 | 0.73 | 0.09 | asserted 1.00 |
| s6 | The server provides session management and model caching, and streams results back per tile. | method 0.97 | 0.71 | 0.03 | asserted 1.00 |
| s7 | The backend is paired with a QGIS plugin that drives tiling, result reassembly, georeferencing, and real-time per-tile status visualization. | method 0.94 | 0.80 | 0.05 | asserted 1.00 |
| s8 | An additional user-interface path injects layer legends as prompts into vision-language models. | method 0.75 | 0.81 | 0.03 | asserted 1.00 |
| s9 | 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. | contribution 0.52 | 0.95 | 0.11 | asserted 1.00 |
| s10 | The inference backend and protocol are open-source and available at https://github.com/microsoft/nxmndr. | contribution 0.89 | 0.53 | 0.04 | asserted 1.00 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
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.
Human review
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.