Skip to content

DeployAI-Environmental-Services/depai-lai

Repository files navigation

Leaf Area Index Estimation for Sentinel-2 Images

This repository provides a deep learning model for Leaf Area Index (LAI estimation) from Sentinel-2 satellite images exposed using a webapp.

Webapp

Input format

The model assumes that the input images are Geotiff images. The minimum allowed size of these images is $500\times500$. 10 Spectral bands are used for the estimation of LAI, which must be in the following order ['B02','B03','B04','B05','B06','B07','B08','B8A','B11','B12']. The input to the model is a list of dictionaries, each dictionary contains the input image path and the offset of the bands for normalization.

[
    {
        "image_path":str,
        "offset":float
    }
]

For more details on the format of the input and output for the model, check model.proto.

Local Development

  • In a terminal, clone the repository
git clone https://github.com/AlbughdadiM/depai-lai.git
  • Go to the repository directory
cd depai-lai
  • If the files model_pb2_grpc.py and model_pb2.py are not there, generate them using
python3.12 -m grpc_tools.protoc -I. --python_out=. --grpc_python_out=. model.proto
  • Build the docker image
docker build . -t lai:v0.1
  • Create a container from the built image
docker run --name=test  -v ./test_folder:/data  --env SHARED_FOLDER_PATH=/data -p 8061:8061 -p 8062:8062 lai:v0.1
  • Run the pytest
pytest test_image_processor.py

Container Registry

  • Generate a personal access token: Github account settings > Developer settings > Personal access tokens (classic). Generate a token with the read:package scope.

  • In a terminal, login to container registry using

docker login ghcr.io -u USERNAME -p PAT
  • Pull the image
docker pull ghcr.io/albughdadim/depai-lai:v0.1
  • Create a container
docker run --name=test  -v ./test_folder:/data  --env SHARED_FOLDER_PATH=/data -p 8061:8061 -p 8062:8062 ghcr.io/albughdadim/depai-lai:v0.1

About

No description or website provided.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages