Weighted Tree-based Crop Classification Models for Imbalanced Datasets

Second place solution in the Zindi AgriFieldNet India Challenge to classify crop types in agricultural fields across Northern India using multispectral observations from Sentinel-2 satellite. Ensembled weighted tree-based models "LGBM, CATBOOST, XGBOOST" with stratified k-fold cross validation, taking advantage of spatial variability around each field within different distances.

MLHub model id: model_ecaas_agrifieldnet_silver_v1. Browse on Radiant MLHub.

ML Model Documentation

Please review the model architecture, license, applicable spatial and temporal extents and other details in the model documentation.

System Requirements

• Git client
• Docker with Compose v1.28 or newer.

Hardware Requirements

Inferencing	Training
30 GB RAM	30 GB RAM

Get Started With Inferencing

First clone this Git repository.

git clone https://github.com/radiantearth/model_ecaas_agrifieldnet_silver.git
cd model_ecaas_agrifieldnet_silver/

After cloning the model repository, you can use the Docker Compose runtime files as described below.

Pull or Build the Docker Image

Pull pre-built image from Docker Hub (recommended):

docker pull docker.io/radiantearth/model_ecaas_agrifieldnet_silver:1

Or build image from source:

docker build -t radiantearth/model_ecaas_agrifieldnet_silver:1 -f Dockerfile .

Run Model to Generate New Inferences

1. Prepare your input and output data folders. The data/ folder in this repository contains some placeholder files to guide you.

   • The data/ folder must contain:
     • input/chips/ Sentinel-2 10m imagery chips for inferencing:
       • images/ Sentinel-2 10m imagery chips for inferencing:
         • Folder name chip_id e.g. 00c23 Sentinel-2 bands 10m:
           • File name: B01.tif Type=Byte, ColorInterp=Coastal
           • File name: B02.tif Type=Byte, ColorInterp=Blue
           • File name: B03.tif Type=Byte, ColorInterp=Green
           • File name: B04.tif Type=Byte, ColorInterp=Red
           • File name: B05.tif Type=Byte, ColorInterp=RedEdge
           • File name: B06.tif Type=Byte, ColorInterp=RedEdge
           • File name: B07.tif Type=Byte, ColorInterp=RedEdge
           • File name: B08.tif Type=Byte, ColorInterp=NIR
           • File name: B8A.tif Type=Byte, ColorInterp=NIR08
           • File name: B09.tif Type=Byte, ColorInterp=NIR09
           • File name: B11.tif Type=Byte, ColorInterp=SWIR16
           • File name: B12.tif Type=Byte, ColorInterp=SWIR22
             • File Format: GeoTIFF, 256x256
             • Coordinate Reference System: WGS84 / UTM
       • fields/ Corresponding field ids for each pixel in Sentinel-2 images:
         • Folder name: chip_id e.g. 00c23 Corresponding field ids:
           • File name: field_ids.tif
           • File Format: GeoTIFF, 256x256
           • Coordinate Reference System: WGS84 / UTM
     • /input/checkpoint/ the model checkpoint lgbms, xgbms, cats. Please note: the model checkpoint is included in this repository.
   • The output/ folder is where the model will write inferencing results.

2. Set INPUT_DATA and OUTPUT_DATA environment variables corresponding with your input and output folders. These commands will vary depending on operating system and command-line shell:

   # change paths to your actual input and output folders
   export INPUT_DATA="/home/my_user/model_ecaas_agrifieldnet_silver/data/input"
   export OUTPUT_DATA="/home/my_user/model_ecaas_agrifieldnet_silver/data/output"

3. Run the appropriate Docker Compose command for your system

   docker-compose up model_ecaas_agrifieldnet_silver_v1
   
   #  If the user is not added to docker group
   sudo -E docker-compose up model_ecaas_agrifieldnet_silver_v1

4. Wait for the docker compose to finish running, then inspect the OUTPUT_DATA folder for results.

Understanding Output Data

Please review the model output format and other technical details in the model documentation.