Testing and training data for machine learning models to detect, classify and count blackbirds damaging agriculture using drone-based imagery

Metadata:

Entity_and_Attribute_Information:

Overview_Description:

Entity_and_Attribute_Overview:

Below you will find a list and description of the files included in this data publication.

VARIABLE DESCRIPTION FILE (1)

1. \Data\_variable_descriptions.csv: Comma-separated values (CSV) file containing a list and description of variables found in the data file. (A description of these variables is also provided in the metadata below.)

Columns include:

Filename = name of data file

Variable = name of variable

Units = units (if applicable)

Precision = precision (if applicable)

Description = description of variable



DATA FILES (1)

1. \Data\EcolInfom_PerformanceDataset.csv: CSV file containing data for evaluating the performance (i.e., accuracy and percent correct) of the Faster-RCNN model used to detect and classify individual blackbirds by species and for red-winged blackbirds (RWBL) sex and age class. This file can be used with the provided R code (\Supplements\Duttenhefner_etal_2025_EcolInform.R) to analyze the results as reported in Duttenhefner et al. (2025).

Variables include:

tile_id = Unique identification for each tile produced for training

background = Prevalent background type occupying the individual tile (sky, green, tan)

total_count = Total number of birds on the tile

unid_accuracy = Model accuracy (%) for detecting individual unidentified birds.

ahy_accuracy = Model accuracy (%) for detecting individual ahy (after hatch year) red-winged blackbirds

hy_accuracy = Model accuracy (%) for detecting individual hy (hatch year) red-winged blackbirds

fem_accuracy = Model accuracy (%) for detecting individual fem (female) red-winged blackbirds

cogr_accuracy = Model accuracy (%) for detecting individual cogr (common grackle)

yhbl_accuracy = Model accuracy (%) for detecting individual yhbl (yellow-headed blackbird)

bhco_accuracy = Model accuracy (%) for detecting individual bhco (brown-headed cowbird)

eust_accuracy = Model accuracy (%) for detecting individual eust (European starling)

unid_correct = Correctly classified (%) individual unidentified birds

ahy_correct = Correctly classified (%) individual ahy (after hatch year) red-winged blackbirds

hy_correct = Correctly classified (%) individual hy (hatch year) red-winged blackbirds

fem_correct = Correctly classified (%) individual fem (female) red-winged blackbirds

cogr_correct = Correctly classified (%) individual cogr (common grackles)

yhbl_correct = Correctly classified (%) individual yhbl (yellow-headed blackbirds)

bhco_correct = Correctly classified (%) individual bhco (brown-headed cowbirds)

eust_correct = Correctly classified (%) individual eust (European starlings)



SUPPLEMENTAL FILES (663)

1. \Supplements\Duttenhefner_etal_2025_EcolInform.R: Text file containing R code. This code can be used with the provided CSV files to replicate results step by step as they appear in Duttenhefner et al. (2025).

2-132. \Supplements\BirdAnnotationXML\Trial [###] - [DESCRIPTION].xml: XML files (131) containing annotations used to train the Faster Region-based Convolutional Neural Network (Faster-RCNN) models to detect individual blackbirds, classify individual blackbirds by species and for RWBL sex and age class, and count blackbirds. These files were created while manually annotating photos with LabelImg (Tzutalin; v1.8.1).

133-263. \Supplements\BirdFrames\Trial [###] - [DESCRIPTION].jpg: Joint Photographic Experts Group (JPG) files (131) containing images used in the Faster Region-based Convolutional Neural Network (Faster-RCNN) models to detect individual blackbirds, classify individual blackbirds by species and for RWBL sex and age class, and count blackbirds.

264-663. \Supplements\FlockFrames\Trial [###] - [DESCRIPTION].jpg: JPG files (400) containing images used in the ResNet-18 convolutional neural network (CNN) model to detect flocks of varying size and distance from the camera.

Entity_and_Attribute_Detail_Citation:

Duttenhefner, Jessica L.; ElSaid, AbdElRahman A.; Klug, Page E. 2025. Machine learning to detect, classify, and count blackbirds damaging agriculture using drone-based imagery: Supporting AI-driven automation for deployment of damage management tools. Ecological Informatics. 92: 103495. https://doi.org/10.1016/j.ecoinf.2025.103495

Metadata_Reference_Information:

Metadata_Date: 20251115

Metadata_Contact:

Contact_Information:

Contact_Organization_Primary:

Contact_Organization: USDA, APHIS, Wildlife Service, National Wildlife Research Center

Contact_Person: Page Klug

Contact_Position: Supervisory Research Wildlife Biologist

Contact_Address:

Address_Type: mailing and physical

Address: 4101 LaPorte Ave.

City: Fort Collins

State_or_Province: CO

Postal_Code: 80521

Country: USA

Contact_Voice_Telephone: 701-630-3776

Contact_Electronic_Mail_Address: page.e.klug@usda.gov

Contact Instructions: This contact information was current as of original publication date. For current information see Contact Us page on: https://doi.org/10.2737/RDS.

Metadata_Standard_Name: FGDC Biological Data Profile of the Content Standard for Digital Geospatial Metadata

Metadata_Standard_Version: FGDC-STD-001.1-1999