samgis 1.5.1

A backend for machine learning instance segmentation on geospatial data even without dedicated graphics cards.

---

title: SamGIS emoji: 🗺️ colorFrom: red colorTo: blue sdk: docker pinned: false license: mit

About this README

I tested these instructions on MacOS, but should work on linux as well.

Segment Anything models

It's possible to prepare the model files using https://github.com/vietanhdev/samexporter/ or using the ones from https://huggingface.co/aletrn/sam-quantized (copy them within the folder /machine_learning_models).

SamGIS - HuggingFace version

The SamGIS HuggingSpace url is https://huggingface.co/spaces/aletrn/samgis. Build the docker image this way:

# clean any old active containers
docker stop $(docker ps -a -q); docker rm $(docker ps -a -q)

# build the base docker image from the repository root folder using ARGs:
# - DEPENDENCY_GROUP=fastapi used by poetry
# VITE__MAP_DESCRIPTION, VITE__SAMGIS_SPACE used by 'docker build'
(
  set -o allexport && source <(cat ./static/.env|grep VITE__) && set +o allexport;
  env|grep VITE__;
  docker build . -f dockerfiles/dockerfile-samgis-base --progress=plain \
  --build-arg DEPENDENCY_GROUP=fastapi \
  --build-arg VITE__MAP_DESCRIPTION=${VITE__MAP_DESCRIPTION} \
  --build-arg VITE__SAMGIS_SPACE=${VITE__SAMGIS_SPACE} \
  --tag registry.gitlab.com/aletrn/gis-prediction
)

# build the image, use the tag "samgis-huggingface"
docker build . --tag registry.gitlab.com/aletrn/samgis-huggingface --progress=plain

Run the container (keep it on background) and show logs

docker run  -d --name samgis-huggingface -p 7860:7860 registry.gitlab.com/aletrn/samgis-huggingface; docker logs -f samgis-huggingface

Test it with curl using a json payload:

URL=http://localhost:7860/infer_samgis
curl -d@./events/payload_point_eolie.json -H 'content-type: application/json' ${URL}

or better visiting the swagger page on http://localhost:7860/docs

SamGIS - lambda AWS version

Build the docker image this way:

# clean any old active containers
docker stop $(docker ps -a -q); docker rm $(docker ps -a -q)

# build the base docker image with the docker aws repository tag
docker build . -f dockerfiles/dockerfile-samgis-base --build-arg DEPENDENCY_GROUP=aws_lambda \
  --tag example-docker-namespace/samgis-base-aws-lambda --progress=plain

# build the final docker image
docker build . -f dockerfiles/dockerfile-lambda-fastsam-api --tag example-docker-namespace/lambda-fastsam-api --progress=plain

Run the container (keep it on background) and show logs

docker run  -d --name lambda-fastsam-api -p 8080:8080 lambda-fastsam-api; docker logs -f lambda-fastsam-api

Test it with curl using a json payload:

URL=http://localhost:8080/2015-03-31/functions/function/invocations
curl -d@./events/payload_point_eolie.json -H 'content-type: application/json' ${URL}

Publish the aws lambda docker image

Login on aws ECR with the correct aws profile (change the example example-docker-namespace/ repository url with the one from the ECR push command instructions page).

Dependencies installation, local execution and local tests

The docker build process needs only the base dependency group plus the aws_lambda or fastapi optional one (right now I use almost only the fastapi version). Install also the test and/or docs groups if needed.

Local execution

If you need to use the SPA frontend follow the frontend instruction here.

You can run the local server using this python command:

uvicorn wrappers.fastapi_wrapper:app --host 127.0.0.1 --port 7860

Change the port and/or the host ip if needed. Test it with curl using a json payload:

URL=http://localhost:7860/infer_samgis
curl -d@./events/payload_point_eolie.json -H 'content-type: application/json' ${URL}

Local execution on MacOS

There is a known issue running the project on MacOS. SamGIS still work also without executing it within a docker container, but is slower during image embedding because of a memory leak caused by CoreML. Here a log about this bug:

[...]
2024-05-15T18:38:37.478698+0200 - INFO - predictors.py - samexporter_predict - (be2506dc-0887-4752-9889-cf12db7501f5) missing instance model mobile_sam, instantiating it now! 
2024-05-15T18:38:37.479582+0200 - INFO - sam_onnx2.py - __init__ - (be2506dc-0887-4752-9889-cf12db7501f5) Available providers for ONNXRuntime: %s 
2024-05-15 18:38:37.673229 [W:onnxruntime:, coreml_execution_provider.cc:81 GetCapability] CoreMLExecutionProvider::GetCapability, number of partitions supported by CoreML: 104 number of nodes in the graph: 566 number of nodes supported by CoreML: 383
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
2024-05-15T18:38:47.691608+0200 - INFO - sam_onnx2.py - __init__ - (be2506dc-0887-4752-9889-cf12db7501f5) encoder_input_name: 
2024-05-15 18:38:47.913677 [W:onnxruntime:, coreml_execution_provider.cc:81 GetCapability] CoreMLExecutionProvider::GetCapability, number of partitions supported by CoreML: 48 number of nodes in the graph: 496 number of nodes supported by CoreML: 221
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
Context leak detected, CoreAnalytics returned false
2024-05-15T18:38:50.926801+0200 - DEBUG - predictors.py - samexporter_predict - (be2506dc-0887-4752-9889-cf12db7501f5) using a mobile_sam instance model... 
[...]

This problem doesn't rise if running it within the docker container.

Tests

Tests are defined in the tests folder in this project. Use PIP to install the test dependencies and run tests.

python -m pytest --cov=samgis --cov-report=term-missing && coverage html

How to update the static documentation with sphinx

This project documentation uses sphinx-apidoc: it's a tool for automatic generation of Sphinx sources that, using the autodoc extension, document a whole package in the style of other automatic API documentation tools. See the documentation page for details. Run the command from the project root:

# missing docs folder (run from project root) initialize this way
# 
cd docs && sphinx-quickstart --project SamGIS --release 1.0.0 --language python --master index

# update docs folder (from project root)
sphinx-apidoc -f -o docs samgis
sphinx-apidoc -f -o docs samgis_core

Then it's possible to generate the HTML pages

cd docs && make html && ../

# to clean old files
cd docs && make clean html && cd ../

The static documentation it's now ready at the path docs/_build/html/index.html.

To create a work in progress openapi json or yaml file use

• extract-openapi-fastapi.py
• extract-openapi-lambda.py (useful to export the json schema request and response from lambda app api)