Analyzing rodent infestations using the geospatial capabilities of Amazon SageMaker

Rodents such as rats and mice are associated with numerous health risks and are known to spread more than 35 diseases. Identifying areas of high rodent activity can help local authorities and pest control organizations effectively plan interventions and eradicate rodents.

In this post, we show how to track and visualize rodent populations using Amazon SageMaker geospatial capabilities. We then visualize the effects of rodent infestations on vegetation and water bodies. Finally, we correlate and visually analyze the number of monkey paste cases reported with rodent sightings in the region. Amazon SageMaker makes it easy for data scientists and machine learning engineers to create, train, and deploy models using geospatial data. The tool makes it easy to access geospatial data sources, perform custom processing operations, use pre-built ML models, and use built-in visualization tools faster and at scale.

notebook

First, we use the Amazon SageMaker Studio notebook with geospatial imagery steps described in Getting Started with Amazon SageMaker Geospatial Capabilities.

Data access

Geospatial Imagery comes pre-installed with SageMaker geospatial capabilities, making it easy to enrich data with geospatial analysis and ML. For our post, we use satellite imagery from Sentinel-2 and rodent activity and monkeypox datasets from the open source NYC Open Data.

First, we use rodent activity and identify the latitude and longitude of rodent sightings and inspections. We then enrich this location information with human-readable street addresses. We create a Vector Enrichment Job (VEJ) in SageMaker Studio Notebook to perform the reverse geocoding operation to convert geographic coordinates (latitude, longitude) to human-readable addresses provided by Amazon Location Service. We create VEJ as follows:

import boto3
import botocore
import sagemaker
import sagemaker_geospatial_map

region = boto3.Session().region_name
session = botocore.session.get_session()
execution_role = sagemaker.get_execution_role()

sg_client= session.create_client(
    service_name="sagemaker-geospatial",
    region_name=region
)
response = sg_client.start_vector_enrichment_job(
    ExecutionRoleArn=execution_role,
    InputConfig=
        'DataSourceConfig': 
            'S3Data': 
                'S3Uri': 's3://<bucket>/sample/rodent.csv'
            
        ,
        'DocumentType': 'CSV'
    ,
    JobConfig=
        "ReverseGeocodingConfig":  
         "XAttributeName": "longitude",
         "YAttributeName": "latitude"
      
    ,
    Name="vej-reversegeo",
)

my_vej_arn = response['Arn']

Visualization of rodent activity in the region

Now we can use SageMaker geospatial capabilities to visualize rodent sightings. Once the VEJ is complete, we run the job output to an Amazon S3 bucket.

sg_client.export_vector_enrichment_job(
    Arn=my_vej_arn,
    ExecutionRoleArn=execution_role,
    OutputConfig=
        'S3Data': 
            'S3Uri': 's3://<bucket>/reversegeo/'
        
    
)

When the export is complete, you’ll see an output CSV file in your Amazon Simple Storage Service (Amazon S3) bucket consisting of your input data (latitude and longitude coordinates) with additional columns: address number, country, label, municipality, neighborhood, zip code, and region of that location appended to the end.

From the output file generated by VEJ, we can use SageMaker geospatial capabilities to output a base map and provide layered visualizations to facilitate collaboration. SageMaker geospatial capabilities are provided with built-in visualization tools provided by Foursquare Studio, which runs natively from the SageMaker notebook via the SageMaker geospatial Map SDK. Below, we can visualize the sightings of rodents and also obtain human-readable addresses for each data point. Address information for each rodent inspection site can be useful for rodent inspection and treatment purposes.

Analyze the impact of rodent infestations on vegetation and water bodies

To analyze the impact of rodent infestation on vegetation and water bodies, we need to divide each site into vegetation, water, and bare ground. Let’s see how we can use these geospatial capabilities to perform this analysis.

SageMaker’s new geospatial capabilities offer easier access to geospatial data such as Sentinel-2 and Landsat 8. Built-in geospatial data access saves weeks of effort, otherwise lost, collecting and processing data from different data providers and suppliers. Also, these geospatial capabilities offer a pre-trained land use land cover (LULC) segmentation model to identify physical material on the Earth’s surface, such as vegetation, water, and bare land.

We use this LULC ML model to analyze the impact of rodent populations on vegetation and water bodies.

In the following code snippet, we first define the coordinates of the region of interest (aoi_coords) in New York. Then we create an Earth Observation Job (EOJ) and select the LULC operation. SageMaker downloads and pre-processes satellite image data for EOJ. Next, SageMaker automatically generates model inference for EOJ. The running time of EOJ varies from a few minutes to hours depending on the number of images processed. You can monitor EOJs using Status get_earth_observation_job Functionality and visualization of EOJ input and output on the map.

aoi_coords = [
    [
            [
              -74.13513011934334,
              40.87856296920188
            ],
            [
              -74.13513011934334,
              40.565792636343616
            ],
            [
              -73.8247144462764,
              40.565792636343616
            ],
            [
              -73.8247144462764,
              40.87856296920188
            ],
            [
              -74.13513011934334,
              40.87856296920188
            ]
    ]
]

eoj_input_config = {
    "RasterDataCollectionQuery": {
        "RasterDataCollectionArn": "arn:aws:sagemaker-geospatial:us-west-2:378778860802:raster-data-collection/public/nmqj48dcu3g7ayw8",
        "AreaOfInterest": 
            "AreaOfInterestGeometry": 
                "PolygonGeometry": 
                    "Coordinates": aoi_coords
                
            
        ,
        "TimeRangeFilter": 
            "StartTime": "2023-01-01T00:00:00Z",
            "EndTime": "2023-02-28T23:59:59Z",
        ,
        "PropertyFilters": 
            "Properties": ["Property": "EoCloudCover": "LowerBound": 0, "UpperBound": 2.0],
            "LogicalOperator": "AND",
        ,
    }
}
eoj_config = 
  "LandCoverSegmentationConfig": 


response = geospatial_client.start_earth_observation_job(
    Name="eoj-rodent-infestation-lulc-example",
    InputConfig=eoj_input_config,
    JobConfig=eoj_config,
    ExecutionRoleArn=execution_role,
)
eoj_arn = response["Arn"]
eoj_arn

Map = sagemaker_geospatial_map.create_map()
Map.set_sagemaker_geospatial_client(sg_client)

Map.render()

time_range_filter = 
    "start_date": "2023-01-01T00:00:00Z",
    "end_date": "2023-02-28T23:59:59Z",



config = "preset": "singleBand", "band_name": "mask"
output_layer = Map.visualize_eoj_output(
    Arn=eoj_arn, config=config, time_range_filter=time_range_filter
)

To visualize rodent populations in relation to vegetation, we overlay rodent population and sighting data on land cover segmentation model predictions. This visualization will help us to determine and analyze the rodent population on vegetation and water bodies.

Visualization of monkeypox incidence and relationship to rodent data

To visualize the relationship between monkeypox cases and rodent sightings, we add the monkeypox database and a geoJSON file for the New York neighborhood boundaries. See the following code:

nybb = pd.read_csv("./nybb.csv")
monkeypox = pd.read_csv("./monkeypox.csv")
dataset = Map.add_dataset(
    "data": nybb
, auto_create_layers=False)
dataset = Map.add_dataset(
    "data": monkeypox
, auto_create_layers=False)

In the SageMaker Studio notebook, we can use the visualization tool created by Foursquare to add layers and charts to the map. Here we have added the monkeypox data as a chart to show the number of monkeypox cases for each district. To show the correlation between monkeypox cases and rodent sightings, we added the borough boundaries as a polygon layer and added a heatmap layer representing rodent activity. The borough boundary layer is colored to match the monkeypox data scheme. As we can see, the borough of Manhattan has the highest concentration of rodent sightings, followed by Brooklyn.

This is supported by simple statistical analysis by calculating the correlation between rodent sightings and monkey occurrences in each township. The calculation yielded an r value of 0.714, implying a positive correlation.

r = np.corrcoef(borough_stats['Concentration (sightings per square km)'], borough_stats['Monkeypox Cases'])

conclusion

In this post, we’ve shown how you can use SageMaker’s geospatial capabilities to get detailed rodent sighting addresses and visualize rodent impacts on vegetation and water bodies. This can help local authorities and pest control organizations to effectively plan interventions and eradicate rodents. We also correlated rodent sightings with monkey occurrences in the area with the built-in visualization tool. With built-in visualization tools using vector enrichment and EOJs, SageMaker geospatial capabilities eliminate the challenges of large-scale geospatial datasets, model training and inference, and provide the ability to quickly explore interactive map projections and geospatial data using 3D graphics acceleration tools.

You can get started with SageMaker geospatial capabilities in two ways:

To learn more, visit Amazon SageMaker Geospatial Capabilities and Getting Started with Amazon SageMaker Geospatial Capabilities. Also, visit our GitHub repo, which has some examples of notebook SageMaker’s geospatial capabilities.

About the authors

Bani Kaushik is a Solutions Architect at AWS. He is passionate about building AI/ML solutions and helping customers innovate on the AWS platform. Outside of work, he enjoys hiking, rock climbing, and swimming.

Clarice Wigal is a Senior Technical Account Manager at AWS focused on helping customers accelerate their cloud adoption journey. Outside of work, Clarice enjoys traveling, hiking, and reading sci-fi thrillers.

Veda Raman is a senior machine learning solutions architect in Maryland. Veda works with customers to help them build efficient, secure and scalable machine learning applications. Veda is interested in helping customers use serverless technologies for machine learning.