How uchihadzoro/deep-learning-super-resolution-mapping works
TRINETRA-SR turns Sentinel-2 imagery into 2.5 m, 10-band outputs with per-pixel uncertainty and radiometric consistency metrics. Its main workflow selects or accepts imagery, runs super-resolution, validates and exports results, then presents them in a browser viewer. The README also documents browser-local inference and an optional server deployment. Browser-engine and viewer relationships below rely on README documentation because their source was not sampled; implementation details beyond that are not inferred.
Main entry points: app.py Optional job API · engine.js Browser engine · app.js Map and result viewer · run_sr.py Scene search and fetch
Components 16
Imagery inputs
- GeoTIFF upload
server/app.py - Scene search and fetch
prototype/run_sr.py
Super-resolution
- Browser engine
docs/engine.js - ONNX model
docs/model/sen2sr_lite.onnx - ONNX model export
prototype/export_onnx.py - SEN2SR-Lite inference
prototype/run_sr.py
Trust and outputs
- Ensemble uncertainty
prototype/run_sr.py - Consistency metrics
prototype/run_sr.py - Assets and GeoTIFFs
prototype/run_sr.py - Demo scene assets
docs/data/index.json
Viewer and deployment
- Map and result viewer
docs/js/app.js - Globe interaction
docs/js/globe.js - Optional job API
server/app.py
Other components
- User
- Sentinel-2 archive
- Sentinel-2 COGs
Connections 22
- User → Map and result viewer: chooses location
README.md - User → Browser engine: uploads imagery
README.md - Map and result viewer → Globe interaction: uses map
README.md - Map and result viewer → Demo scene assets: shows samples
README.md - Browser engine → Sentinel-2 archive: searches scenes
README.md - Browser engine → Sentinel-2 COGs: reads windows
README.md - Browser engine → ONNX model: runs inference
README.md - Browser engine → Map and result viewer: provides results
README.md - ONNX model export → SEN2SR-Lite inference: loads model
prototype/export_onnx.py - ONNX model export → ONNX model: exports model
README.md - Optional job API → Scene search and fetch: fetches scenes
server/app.py - Optional job API → SEN2SR-Lite inference: runs inference
server/app.py - Optional job API → Ensemble uncertainty: computes ensemble
server/app.py - Optional job API → Consistency metrics: computes metrics
server/app.py - Optional job API → Assets and GeoTIFFs: exports results
server/app.py - Scene search and fetch → Sentinel-2 archive: searches scenes
prototype/run_sr.py - Scene search and fetch → Sentinel-2 COGs: reads bands
prototype/run_sr.py - SEN2SR-Lite inference → Ensemble uncertainty: ensembles passes
prototype/run_sr.py - Assets and GeoTIFFs → Consistency metrics: computes metrics
prototype/run_sr.py - Assets and GeoTIFFs → Demo scene assets: writes demo assets
prototype/run_sr.py - User → Optional job API: requests server run
README.md - GeoTIFF upload → Optional job API: submits GeoTIFF
server/app.py