MapSR: Prompt-Driven Land Cover Map Super-Resolution via Vision Foundation Models

2026-04-16 · Source: Computer Vision and Pattern Recognition · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Geospatial AI · Depth: Expert, quick

Summary

MapSR is a novel prompt-driven framework designed for land-cover map super-resolution, addressing the high cost associated with dense high-resolution (HR) annotations. This framework enhances coarse low-resolution (LR) land-cover products into HR maps at the resolution of input imagery. Unlike existing weakly supervised methods that retrain dense predictors, MapSR decouples supervision from model training. It extracts class prompts from frozen vision foundation model features using a lightweight linear probe, leveraging LR labels only once. High-resolution mapping then proceeds via training-free metric inference and graph-based prediction refinement. On the Chesapeake Bay dataset, MapSR achieved 59.64% mIoU without any HR labels, competing with strong weakly supervised baselines and outperforming a fully supervised baseline. Crucially, MapSR reduces trainable parameters by four orders of magnitude and shortens training time from hours to minutes.

Key takeaway

For AI Scientists and Computer Vision Engineers developing land-cover mapping solutions, MapSR offers a compelling approach to achieve high-resolution outputs with significantly reduced annotation and computational burdens. You should consider integrating prompt-driven frameworks like MapSR to accelerate development cycles and deploy scalable HR mapping under tight budget constraints, especially when dense HR labels are prohibitively expensive or unavailable.

Key insights

MapSR uses prompt-driven vision foundation models for efficient, training-free land-cover super-resolution with minimal annotation.

Principles

• Decouple supervision from model training
• Leverage frozen foundation model features
• Employ graph-based spatial refinement

Method

MapSR extracts class prompts from frozen vision foundation model features via a linear probe, then performs HR mapping using training-free metric inference and graph-based prediction refinement.

In practice

• Apply to land-cover mapping
• Reduce annotation costs
• Accelerate model training

Topics

• Map Super-Resolution
• Vision Foundation Models
• Land Cover Mapping
• Prompt-Driven Learning
• Graph-Based Refinement

Code references

• rikirikirikiriki/MapSR

Best for: AI Scientist, Computer Vision Engineer, Research Scientist

Related on AIssential

• MapSR: Prompt-Driven Land Cover Map Super-Resolution via Vision Foundation Models — MapSR enables high-resolution land-cover mapping with minimal annotation and compute by decoupling supervision from training.
• AI Maps 13 Million Buildings in One of the World’s Most Remote Regions — AI-powered satellite imagery can accurately map millions of buildings in remote, challenging terrains.
• Scalable Deep Learning Framework for Global High-Resolution Land Use Reconstruction — AI4Land uses U-Net on HPC to reconstruct global 1 km land use from coarse data, improving climate model inputs.
• Training the AIs' Eyes: How Roboflow is Making the Real World Programmable, with CEO Joseph Nelson — Computer vision, though trailing language models, is approaching its "ChatGPT moment" through specialized, efficient, and edge-deployable models.
• GMBFormer: An NDVI-Guided Global Memory Bank Transformer for Urban Green-Space Extraction from Ultra-High-Resolution Imagery — GMBFormer improves urban green-space extraction by decoupling NDVI and using a global memory bank for semantic reuse.
• A Scale-Adaptive Framework for Joint Spatiotemporal Super-Resolution with Diffusion Models — A scale-adaptive framework reuses one architecture for spatiotemporal super-resolution across diverse scaling factors.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computer Vision and Pattern Recognition.