imageomics/kabr-behavior-telemetry

Dataset Card for KABR Behavior Telemetry

Synchronized frame-level telemetry, detections, and behavior annotations from drone wildlife monitoring in Kenya, enabling research on animal behavior analysis and optimal drone survey protocols.

Dataset Details

Dataset Description

<<<<<<< HEAD - Curated by: Jenna M. Kline, Elizabeth Campolongo - Language(s): English (metadata and documentation) - Homepage: KABR Project - Repository: kabr-behavior-telemetry - Paper: In preparation

• Curated by: Jenna M. Kline, Elizabeth Campolongo, Matt Thompson, Maksim Kholiavchenko, Otto Brookes
• Language(s): English (metadata and documentation)
• Homepage: KABR Project
• Repository: kabr-behavior-telemetry
• Paper: Integrating Biological Data into Autonomous Remote Sensing Systems

  6b503d0331e742072a14db5d7a7dc6bc465cc15d

This dataset provides frame-level integration of drone telemetry (GPS position, altitude, camera settings), animal detection bounding boxes, and expert-annotated behaviors from aerial wildlife monitoring in Kenyan savannas. Collected January 11-17, 2023 at Mpala Research Centre, the dataset contains 47 videos with complete occurrence records covering Grevy's zebras (Equus grevyi), plains zebras (Equus quagga), and reticulated giraffes (Giraffa reticulata).

The dataset was developed to analyze optimal drone flight parameters for wildlife behavior research—correlating altitude, speed, and camera settings with data quality and animal disturbance levels. It implements Darwin Core biodiversity standards with Humboldt Eco extensions for ecological inventory data, ensuring interoperability with biodiversity databases like GBIF.

Key features:

• 47 complete video occurrence files with ~10,000-66,000 frames each
• 68 video-level Darwin Core events with GPS bounds and temporal coverage
• 17 session-level events aggregating mission-level metadata
• Frame-synchronized data: GPS coordinates, camera EXIF, detection boxes, behavior labels
• Behavior ethogram: Walking, running, grazing, vigilance, social interactions, and more
• Multi-species coverage: Three focal species across diverse habitats

Supported Tasks and Applications

This dataset supports computer vision, ecological analysis, and autonomous systems research:

🤖 Computer Vision Tasks:

• Object Detection (bounding boxes around animals)
• Multi-Object Tracking (ID consistency across frames in mini-scenes)
• Behavior Recognition (activity classification)
• Scale-Invariant Detection (animals at varying altitudes/distances)

🌿 Ecological Applications:

• Behavioral time budgets and activity patterns
• Habitat use analysis
• Group size and composition estimation
• Flight parameter impact on data quality
• Animal response to drone presence

🚁 Drone Systems Research:

• Optimal altitude/speed/distance determination
• Camera parameter optimization for wildlife
• Detection performance vs. flight parameters
• Disturbance minimization protocols

Dataset Structure

Directory Organization

kabr-behavior-telemetry/
├── data/
│   ├── occurrences/           # Frame-level occurrence records (47 videos)
│   │   ├── 11_01_23-DJI_0977.csv
│   │   ├── 11_01_23-DJI_0978.csv
│   │   └── ...
│   ├── video_events.csv       # Darwin Core Event records (68 videos)
│   └── session_events.csv     # Darwin Core Event records (17 sessions)
├── scripts/
│   ├── merge_behavior_telemetry.py    # Generate occurrence files
│   ├── update_video_events.py         # Add annotation file paths
│   ├── add_event_times.py             # Extract temporal bounds
│   └── add_gps_data.py                # Extract GPS statistics
├── metadata/
│   ├── DATA_DICTIONARY.md             # Complete field descriptions
│   └── event_session_fields.csv       # Field metadata
└── README.md

Data Instances

Occurrence Files (data/occurrences/*.csv):

Each CSV contains frame-by-frame records for one video. Example from 11_01_23-DJI_0977.csv:

Field	Example Value	Description
date	11_01_23	Recording date
video_id	DJI_0977	Video identifier
frame	0	Frame number
date_time	2023-01-11 16:04:03,114,286	Timestamp with μs precision
latitude	0.399770	GPS latitude (WGS84)
longitude	36.891217	GPS longitude (WGS84)
altitude	20.2	Altitude (m above sea level)
iso	100	Camera ISO
xtl, ytl, xbr, ybr	1245, 678, 1389, 842	Bounding box coordinates
id	12	Mini-scene/track ID
behaviour	walking	Behavior class

Naming Convention:

{date}_{video_id}.csv
Example: 11_01_23-DJI_0977.csv
         └─date─┘ └video_id┘

Temporal Information:

• Date range: 2023-01-11 to 2023-01-17
• Time coverage: Morning (09:38) to afternoon (16:28)
• Dry season in Laikipia, Kenya

Data Fields

See metadata/DATA_DICTIONARY.md for complete field descriptions.

Key field groups:

🌿 Darwin Core Event Fields (video_events.csv, session_events.csv):

• Event identifiers and temporal coverage
• Geographic coordinates and bounds
• Sampling protocol descriptions
• Taxonomic scope and abundance
• Humboldt Eco extensions for inventory data

📍 Geolocation (occurrence files):

• GPS latitude/longitude (WGS84)
• Altitude above sea level
• Launch point and bounding box (event files)

📷 Camera Metadata (occurrence files):

• ISO, shutter speed, aperture
• Focal length and zoom ratio
• Color temperature and mode

🦓 Detection Annotations (occurrence files):

• Bounding box coordinates (CVAT format)
• Track ID for multi-frame sequences
• Occlusion and truncation flags

🏃 Behavior Labels (occurrence files):

• Activity classification (ethogram-based)
• Behavioral state at frame capture
• Mini-scene grouping

Data Splits

This dataset does not include pre-defined train/val/test splits. Recommended splitting strategies:

Temporal Split:

• Train: Jan 11-13 (sessions 1-8)
• Val: Jan 16 (session 9)
• Test: Jan 17 (sessions 10-11)

Spatial Split:

• Split by location clusters based on GPS coordinates

Species-Stratified:

• Ensure all three species in each split

Mission-Based:

• Keep complete sessions together (do not split individual videos)

Platform and Mission Specifications

🚁 Platform Details

Type: UAV (Unmanned Aerial Vehicle)

Hardware:

• Primary Platform: DJI Mavic Air 2
• Secondary Platform: DJI Mini
• Max flight time: ~25-30 minutes
• Wind resistance: Moderate (class 5 winds, ~8-10 m/s)

Autonomy:

• Mode: Manual flight with GPS stabilization
• Navigation: Operator-controlled following focal groups
• Collision avoidance: Obstacle detection enabled
• Return-to-home: Automatic on signal loss

📷 Sensor Specifications

Primary Sensor: DJI Integrated Camera

• Type: RGB
• Resolution: 5472 × 3078 pixels (5.4K)
• Frame rate: 24-30 fps
• Bit depth: 8-bit
• Format: MP4 video

Telemetry Included:

• GPS coordinates (SRT files embedded in video)
• IMU data (altitude, orientation)
• Camera settings (ISO, shutter, aperture, focal length)
• Timestamp synchronization

🗺️ Mission Parameters

Flight Specifications:

• Altitude: 20-50 m AGL (above ground level)
• Typical altitude: 30-40 m
• Speed: 0-5 m/s (adaptive to animal movement)
• Flight pattern: Focal animal follows (manual tracking)
• Duration per mission: 5-50 minutes

Environmental Conditions:

• Season: Dry season (January)
• Weather: Clear to partly cloudy
• Wind: Light to moderate
• Time of day: Morning (09:00-12:00) and afternoon (14:00-16:30)

🔍 Sampling Protocol

Survey Design:

• Focal group follows: Single herd tracked continuously per session
• Opportunistic sampling of observed groups
• Continuous video recording during follows

Flight Operations:

• Licensed drone operators with Kenya Civil Aviation Authority approval
• Maintained minimum altitude of 20m to minimize disturbance
• Animals monitored for behavioral response; flight aborted if disturbance detected

Data Collection:

• Continuous video recording at 5.4K resolution
• GPS telemetry embedded in SRT sidecar files
• Frame extraction in CVAT for annotation

Quality Control:

• Field notes recorded for each session
• Video quality assessed before annotation
• Behavior annotations reviewed by expert ecologists

Dataset Creation

Curation Rationale

This dataset was created to address two key research questions:

1. What drone flight parameters optimize behavioral data quality? By correlating altitude, speed, distance, and camera settings with annotation completeness and animal visibility, researchers can develop evidence-based protocols for wildlife monitoring.

2. Can we quantify animal disturbance from drone presence? Frame-level behavior annotations allow detection of alert, fleeing, or disrupted behaviors that indicate drone impact.

The dataset fills a critical gap: while many drone wildlife datasets provide detection boxes, few include detailed behavior labels synchronized with flight telemetry. This enables research on the trade-offs between data quality and animal welfare.

Source Data

Data Collection and Processing

Field Collection:

1. Planning:

   • Sites selected based on known zebra and giraffe populations at Mpala Research Centre
   • Flights conducted during peak activity hours (morning/afternoon)
   • Safety briefings and airspace clearance for each flight

2. Collection:

   • Operators located focal groups via binoculars or vehicle sighting
   • Drones launched 50-100m from animals
   • Continuous video recording while following group movements
   • Flight logs automatically recorded in SRT files
   • Field notes on weather, behavior, and technical issues

3. Post-Processing:

   • Videos transferred from SD cards with immediate backup
   • SRT files extracted for telemetry data
   • Frame extraction at 1 fps in CVAT annotation platform
   • Detection bounding boxes drawn for all visible animals
   • Mini-scenes identified (continuous behavioral sequences)
   • Behavior labels applied by trained ecologists
   • Quality review of all annotations

Software and Tools Used:

• Flight planning: DJI Fly app
• Video capture: DJI Mavic Air 2 / DJI Mini onboard recording
• Frame extraction: CVAT (Computer Vision Annotation Tool)
• Annotation: CVAT with custom behavior taxonomy
• Telemetry parsing: Custom Python scripts
• Data merging: merge_behavior_telemetry.py (this repository)

Who are the source data producers?

Field Team:

• Jenna M. Kline (Ohio State University) - Drone operations, field coordination
• Elizabeth Campolongo (Rensselaer Polytechnic Institute) - Drone operations, data collection
• Matt Thompson (Ohio State University) - Drone operations, field support
• Local field assistants from Mpala Research Centre

Local Collaboration:

• Mpala Research Centre provided logistical support and site access
• Kenya Wildlife Service granted research permits
• Local communities consulted on flight operations

Annotations

Annotation Process

🤖 Annotation Method:

• Semi-automated (CVAT tracking tools + manual review and behavior labeling)

Tools Used:

• Software: CVAT (Computer Vision Annotation Tool)
• Version: Web-based platform (2023)

Annotation Guidelines:

• All visible animals annotated with bounding boxes
• Bounding boxes drawn tightly around animal body
• Partial occlusions included if >30% of animal visible
• Track IDs maintained across frames within mini-scenes
• Behavior labels applied based on dominant activity in frame
• Uncertain behaviors marked for expert review

Quality Control:

• Annotator training: 4 hours on example videos with expert feedback
• Inter-annotator agreement: Not formally quantified (small expert team)
• Review process: Senior ecologist (Kline) reviewed 100% of behavior labels
• Difficult cases: Discussed in team meetings, consensus labels applied
• Annotation confidence: Not explicitly scored

Annotation Coverage:

• Fully annotated: No (not all frames have animals)
• Frames with visible animals: ~90% annotated
• Behavior labels: Applied to mini-scenes (continuous sequences)
• Missing annotations: Frames without animals or poor quality (blur, occlusion)

Who are the annotators?

Annotator Team:

• Number of annotators: 3 primary, 2 reviewers
• Expertise: Graduate students in ecology/computer science with wildlife identification training
• Training provided: 4 hours initial training + ongoing feedback
• Compensation: Academic credit and authorship

Subject Matter Experts:

• Daniel Rubenstein - Guidance on zebra and giraffe behavior
• Charles Stewart - Computer vision and annotation protocols
• Tanya Berger-Wolf - Funding, project oversight
• Elizabeth Campolongo - Data science and annotation review
• Matt Thompson - Software development and data processing
• Jenna Kline - Drone operations, project lead, annotation review

Personal and Sensitive Information

⚠️ Privacy and Security Considerations:

Human Subjects:

• No humans visible in imagery
• Note: Flights conducted in remote areas away from settlements

Endangered Species:

• Contains endangered/threatened species: Equus grevyi (Grevy's zebra, Endangered)
• Location precision: Full GPS coordinates included (site is within protected research center)
• Consultation: Mpala Research Centre and Kenya Wildlife Service approved data sharing

Cultural Sensitivity:

• Traditional lands: Mpala Research Centre operates with community consent

Security:

• No security concerns
• Data collected in collaboration with local authorities
• Full coordinates shared to support scientific use

Considerations for Using the Data

Dataset Statistics

Species Distribution:

Species (Scientific Name)	Common Name	Videos	Sessions	Individuals (range)
Equus grevyi	Grevy's zebra	5	3	3-7
Equus quagga	Plains zebra	30	11	2-12
Giraffa reticulata	Reticulated giraffe	6	2	4-8
Mixed	Multiple species	6	1	2-4

Class Balance:

• Plains zebra over-represented (opportunistic sampling)
• Grevy's zebra limited by lower population density
• Giraffes limited to specific habitat types

Video Characteristics:

• Frame count range: 10,000-66,000 frames per video
• Duration range: 3-50 minutes per video
• Altitude range: 8-72 m above sea level
• Typical animal size in frame: 50-200 pixels (height)

Behavior Distribution:

• Walking: ~40%
• Grazing: ~25%
• Standing/vigilance: ~20%
• Running: ~10%
• Other (social, nursing, etc.): ~5%

Bias, Risks, and Limitations

⚠️ Known Biases:

1. Geographic Bias:

   • Data from single site (Mpala Research Centre, Laikipia)
   • May not generalize to other savanna ecosystems
   • Represents dry season only, captured during drought conditions

2. Temporal Bias:

   • Morning and afternoon flights only (battery/weather constraints)
   • Nocturnal or dawn/dusk behavior not captured
   • Single month snapshot (seasonal variation not represented)

3. Species Bias:

   • Plains zebra over-represented (most abundant species)
   • Grevy's zebra limited by population size
   • No data on smaller species (<50 cm body size)

4. Environmental Bias:

   • Dry season habitat conditions
   • Drought-affected vegetation
   • Clear to partly cloudy weather only
   • No wet season or dense vegetation scenarios

5. Detection Bias:

   • Animals in open areas more likely to be followed
   • Dense vegetation reduces detection probability
   • Cryptic species under-represented

Technical Limitations:

• Image Quality: Variable due to altitude, lighting, and atmospheric conditions
• Coverage Gaps: 21 videos lack occurrence data due to missing/corrupted SRT files or failed processing
• Annotation Limitations: Behavior labels are subjective; inter-observer agreement not quantified
• GPS Accuracy: ±5-10m typical; may drift during long flights

Ethical Limitations:

• Animal Welfare: Potential for disturbance despite mitigation efforts
• Data Usage: GPS coordinates could theoretically enable harmful uses (though species are common and well-protected at Mpala)

Recommendations

Best Practices for Using This Dataset:

1. For Detection/Tracking Models:

   • Account for altitude-dependent scale variation (20-50m range)
   • Consider species-specific detection difficulty (giraffes easier than zebras)
   • Test generalization to new sites (single-location training data)

2. For Behavior Recognition:

   • Class imbalance exists; consider weighted loss or resampling
   • Behavior labels are coarse; fine-grained states may be ambiguous
   • Temporal context improves accuracy (behaviors occur in sequences)

3. For Ecological Analysis:

   • Do not extrapolate to wet season without additional data
   • Account for detection probability varying by habitat/altitude
   • Animal counts are minimum estimates (some individuals may be hidden)

4. For Drone Protocol Development:

   • Correlate altitude/speed with detection rate and annotation completeness
   • Monitor for behavioral responses in data (alert, flee behaviors)
   • Consider trade-offs between data quality and disturbance risk

Ethical Use:

• Do not use for unethical wildlife targeting or harassment
• Respect that full GPS coordinates enable site replication for conservation research
• Cite dataset appropriately and acknowledge indigenous land stewardship

What This Dataset Should NOT Be Used For:

• Estimating absolute population sizes (sampling is not systematic)
• Generalizing to wet season, nighttime, or other habitats/regions

Licensing Information

Dataset License: CC0 1.0 Universal (CC0 1.0) Public Domain Dedication

Citation Requirement: While CC0 does not legally require citation, we strongly request that you cite the dataset and associated paper if you use this data (see Citation section).

Code License: MIT License for scripts in this repository

Citation

If you use this dataset, please cite:

Dataset:

@misc{kline2024kabr_telemetry,
  author = {Kline, Jenna M. and Campolongo, Elizabeth and Thompson, Matt and
            Kholiavchenko, Maksim and Brookes, Otto and Berger-Wolf, Tanya and
            Stewart, Charles V. and Stewart, Christopher},
  title = {KABR Behavior Telemetry: Frame-Level Drone Wildlife Monitoring Dataset},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/imageomics/kabr-behavior-telemetry}
}

Associated Paper:

@article{kline2024integrating,
  title = {Integrating Biological Data into Autonomous Remote Sensing Systems for
           In Situ Imageomics: A Case Study for Kenyan Animal Behavior Sensing with
           Unmanned Aerial Vehicles (UAVs)},
  author = {Kline, Jenna M. and Campolongo, Elizabeth and Thompson, Matt and others},
  journal = {arXiv preprint arXiv:2407.16864},
  year = {2024},
  url = {https://arxiv.org/abs/2407.16864}
}

FAIR² Drone Data Standard:

@article{kline2025fair2,
  title = {Toward a FAIR² Standard for Drone-Based Wildlife Monitoring Datasets},
  author = {Kline, Jenna and others},
  year = {2025},
  note = {In preparation}
}

Acknowledgements

This work was supported by the Imageomics Institute, which is funded by the US National Science Foundation's Harnessing the Data Revolution (HDR) program under Award #2118240 (Imageomics: A New Frontier of Biological Information Powered by Knowledge-Guided Machine Learning). Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

We thank:

• Mpala Research Centre and Jackson Miliko for logistical support and site access
• Kenya Wildlife Service for research permits
• Kenya Civil Aviation Authority for drone operation clearances
• Local field assistants from Mpala Research Centre
• Annotation team: Maksim Kholiavchenko (Rensselaer Polytechnic Institute) - ORCID: 0000-0001-6757-1957 Jenna Kline (The Ohio State University) - ORCID: 0009-0006-7301-5774 Michelle Ramirez (The Ohio State University) Sam Stevens (The Ohio State University) Alec Sheets (The Ohio State University) - ORCID: 0000-0002-3737-1484 Reshma Ramesh Babu (The Ohio State University) - ORCID: 0000-0002-2517-5347 Namrata Banerji (The Ohio State University) - ORCID: 0000-0001-6813-0010 Elizabeth Campolongo (Imageomics Institute, The Ohio State University) - ORCID: 0000-0003-0846-2413 Matthew Thompson (Imageomics Institute, The Ohio State University) - ORCID: 0000-0003-0583-8585 Nina Van Tiel (Eidgenössische Technische Hochschule Zürich) - ORCID: 0000-0001-6393-5629

Conservation Partners:

• Mpala Research Centre, Laikipia County, Kenya
• Grevy's Zebra Trust

Data Collection Permits:

• Kenya Wildlife Service research permit
• Kenya Civil Aviation Authority drone operations clearance
• Nacosti research license

Validation and Quality Metrics

🤖 AI-Readiness Validation:

• Machine-readable metadata (YAML front matter complete)
• Structured annotations in Darwin Core format
• Train/val/test splits pre-defined (users should create)
• Class distribution documented
• Data loading code provided (Python scripts)
• Example notebooks (planned)

🌿 Darwin Core Validation:

• Event records complete and valid
• Occurrence records complete and valid (47/68 videos)
• Scientific names validated against GBIF backbone
• Coordinates in WGS84
• Sampling protocol documented
• GBIF dataset registration (planned)

⚠️ FAIR² Compliance Checklist:

• Findable: DOI to be assigned
• Accessible: Open access via GitHub/Hugging Face
• Interoperable: Darwin Core, Humboldt Eco, CSV/JSON formats
• Reusable: CC0 license, full provenance documented
• AI-Ready: Machine-readable, structured, versioned

Code and Tools

Data Loading (Python):

import pandas as pd

# Load session-level events
sessions = pd.read_csv('data/session_events.csv')

# Load video-level events
videos = pd.read_csv('data/video_events.csv')

# Load occurrence records for a specific video
occurrences = pd.read_csv('data/occurrences/11_01_23-DJI_0977.csv')

# Filter to frames with detections
detections = occurrences.dropna(subset=['xtl', 'ytl', 'xbr', 'ybr'])

# Group by behavior
behavior_counts = detections.groupby('behaviour').size()

Visualization Example:

import matplotlib.pyplot as plt
import geopandas as gpd
from shapely.wkt import loads

# Plot session footprints
sessions_with_gps = sessions.dropna(subset=['footprintWKT'])
geometries = [loads(wkt) for wkt in sessions_with_gps['footprintWKT']]
gdf = gpd.GeoDataFrame(sessions_with_gps, geometry=geometries, crs='EPSG:4326')

fig, ax = plt.subplots(figsize=(10, 10))
gdf.plot(ax=ax, alpha=0.5, edgecolor='black')
plt.title('Session Geographic Coverage')
plt.xlabel('Longitude')
plt.ylabel('Latitude')
plt.show()

Processing Scripts:

See scripts/ directory for:

• merge_behavior_telemetry.py - Generate occurrence files from source data
• update_video_events.py - Add annotation file references
• add_event_times.py - Extract temporal bounds
• add_gps_data.py - Calculate GPS statistics

Glossary

• AGL: Above Ground Level - altitude measured from terrain surface
• Darwin Core: Biodiversity data standard maintained by TDWG
• Ethogram: Catalog of behaviors exhibited by a species
• FAIR²: FAIR principles extended for AI-ready datasets
• Humboldt Eco: Extension of Darwin Core for ecological inventory data
• Mini-scene: Continuous behavioral sequence tracked across frames
• Occurrence: Darwin Core term for species observation record
• SRT: SubRip subtitle format; used for drone telemetry embedding
• TDWG: Biodiversity Information Standards (Taxonomic Databases Working Group)
• UAV: Unmanned Aerial Vehicle (drone)
• WKT: Well-Known Text format for geographic geometries

Dataset Card Authors

Jenna M. Kline, Elizabeth Campolongo, Matt Thompson

Dataset Card Contact

For questions about this dataset:

• Primary Contact: Jenna M. Kline ([email protected])
• Issues: GitHub repository issues
• KABR Project: https://imageomics.github.io/KABR/

---

Version History:

• v1.1.0 (2026-01-02): Added occurrence files, GPS data, temporal bounds, updated Darwin Core events
• v1.0.0 (2024-12-31): Initial release with session and video event metadata

---

This dataset card follows the FAIR² Drone Data Standard and extends the Imageomics dataset card template.