{"success":true,"data":{"author":"Imageomics Institute (Kline J. M., Campolongo E.)","bucketAccessType":"public","bucketName":"imageomics/kabr-behavior-telemetry","bucketPrefix":"","category":"Life Sciences","cloudProvider":"huggingface","createdAt":"2026-11-16T02:38:32.312Z","description":"Frame-synchronised drone telemetry, animal detection boxes and expert behaviour annotations from aerial monitoring of Grevy's zebras (Equus grevyi), plains zebras (Equus quagga) and reticulated giraffes (Giraffa reticulata) at the Mpala Research Centre, Kenya, 11 to 17 January 2023. 57 videos each have a complete occurrence file of 10,000 to 66,000 frames joining GPS position, altitude, speed and camera EXIF with bounding boxes and behaviour labels from an ethogram (walking, running, grazing, vigilance, social interactions). Records follow Darwin Core with the Humboldt Eco extension, with 68 video-level and 18 session-level events, so they can be exchanged with GBIF-style biodiversity tools. The dataset was built to ask which flight altitudes, speeds and camera settings give usable behaviour data with least disturbance to animals. Measured: 62 CSV files, about 1.6 GB, under data/occurrences and event tables. Load with pandas. Licensed CC0 1.0 by the Imageomics Institute; citation of the dataset is requested.","displayTitle":"KABR Behavior Telemetry: Frame-Level Drone Telemetry, Detections and Behaviour Labels (Darwin Core)","id":"85346d64-a990-51f3-80b9-e199ca3684ba","license":"CC0-1.0","licenseEvidenceUrl":"https://huggingface.co/datasets/imageomics/kabr-behavior-telemetry","objectCountEstimate":65,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"imageomics/kabr-behavior-telemetry\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"imageomics/kabr-behavior-telemetry\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=imageomics, license:cc0-1.0, size_categories:100K<n<1M, downloads=309"],"format":"mixed","generatedAt":"2026-11-16T02:38:36.689Z","layout":{"example":"data/consolidated_metadata.csv","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"frame,id,iso,shutter,fnum,ev,ct,color_md,focal_len,dzoom_ratio,latitude,longitude,altitude,outside_x,occluded_x,keyframe_x,xtl,ytl,xbr,ybr,z_order_x,label,source,keyframe_y,outside_y,occluded_y,points,z_order_y,behaviour,video\n0,2.0,100.0,1/1600.0,280.0,0.0,5049.0,default,224.0,\"10000, delta:0\",0.381851,36.863116,16.5,0.0,0.0,1.0,742.0,1232.0,833.0,1298.0,0.0,Zebra,manual,1.0,0.0,0.0,\"19.89,17.36\",0.0,Walk,12_01_23-DJI_0994\n0,1.0,100.0,1/1600.0,280.0,0.0,5049.0,default,224.0,\"10000, delta:0\",0.381851,36.863","encoding":"utf8","key":"data/consolidated_metadata.csv","truncatedAt":512}],"sampleObjects":["data/consolidated_metadata.csv","data/occurrences/11_01_23-DJI_0977.csv","data/occurrences/11_01_23-DJI_0978.csv","data/occurrences/11_01_23-DJI_0979.csv","data/occurrences/11_01_23-DJI_0980.csv"],"sidecars":[{"bytes":"---\nlicense: cc0-1.0\nlanguage:\n- en\npretty_name: KABR Behavior Telemetry - FAIR² Drone Wildlife Monitoring Dataset\n\ntask_categories:\n  - object-detection\n  - video-classification\n\ntags:\n  - biology\n  - ecology\n  - wildlife-monitoring\n  - drone\n  - uav\n  - aerial-imagery\n  - animal-behavior\n  - zebra\n  - giraffe\n  - kenya\n  - savanna\n  - fair-drones\n\nsize_categories:\n  - 100K<n<1M\n\ndescription: Frame-level behavior telemetry dataset combining drone GPS tracks, camera metadata, animal detections, and behavior annotations from wildlife monitoring in Kenya. Supports detection, tracking, and behavioral analysis research.\n\n# FAIR² COMPLIANCE METADATA\nfair2_compliance:\n  findable:\n    doi: \"\"  # To be assigned\n    metadata_registry: [\"DataCite\", \"GBIF\"]\n  accessible:\n    open_access: true\n    authentication_required: false\n  interoperable:\n    standards: [\"Darwin Core\", \"TDWG\", \"Humboldt Eco\", \"FAIR2\"]\n  reusable:\n    license_clear: true\n    provenance_documented: true\n  ai_ready:\n    machine_readable: true\n    structured_annotations: true\n\n# DARWIN CORE COMPLIANCE\ndarwin_core:\n  event_coverage:\n    start_date: \"2023-01-11\"\n    end_date: \"2023-01-17\"\n    decimal_latitude: 0.36\n    decimal_longitude: 36.88\n    coordinate_uncertainty_meters: 10\n    locality: \"Mpala Research Centre, Laikipia County, Kenya\"\n    habitat: \"Open savanna and bushy woodland\"\n\n  occurrence_info:\n    kingdom: \"Animalia\"\n    taxa_included: [\"Equus grevyi\", \"Equus quagga\", \"Giraffa reticulata\"]\n    sampling_protocol: \"Aerial drone video survey at 20-50m altitude with continuous recording and frame-by-frame behavior annotation\"\n\n# PLATFORM SPECIFICATIONS\nplatform:\n  type: \"UAV\"\n  manufacturer: \"DJI\"\n  model: \"Mavic Air 2, DJI Mini\"\n  autonomy_mode: \"manual\"\n\n# SENSOR SPECIFICATIONS\nsensors:\n  - type: \"RGB\"\n    manufacturer: \"DJI\"\n    model: \"Integrated camera\"\n    resolution: [5472, 3078]\n\n# MISSION PARAMETERS\nmission:\n  altitude_m: 35\n  speed_ms: 3\n  telemetry_available: true\n---\n\n# Dataset Card for KABR Behavior Telemetry\n\n**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.**\n\n## Dataset Details\n\n### Dataset Description\n\n- **Curated by:** Jenna M. Kline, Elizabeth Campolongo\n- **Language(s):** English (metadata and documentation)\n- **Homepage:** [KABR Project](https://imageomics.github.io/KABR/)\n- **Repository:** [kabr-behavior-telemetry](https://github.com/Imageomics/fair_drones/tree/main/examples/kabr)\n- **Paper:** In preparation \n\nThis 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 57 videos with complete occurrence records covering Grevy's zebras (*Equus grevyi*), plains zebras (*Equus quagga*), and reticulated giraffes (*Giraffa reticulata*).\n\nThe 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.\n\nKey features:\n- **57 complete video occurrence files** with ~10,000-66,000 frames each\n- **68 video-level Darwin Core events** with GPS bounds and temporal coverage\n- **18 session-level events** aggregating mission-level metadata\n- **Frame-synchronized data**: GPS coordinates, camera EXIF, detection boxes, behavior labels\n- **Behavior ethogram**: Walking, running, grazing, vigilance, social interactions, and more\n- **Multi-species coverage**: Three focal species across diverse habitats\n\n### Supported Tasks and Applications\n\nThis dataset supports computer vision, ecological analysis, and autonomous systems research:\n\n**🤖 Computer Vision Tasks:**\n- Objec","encoding":"utf8","key":"README.md","name":"README.md","size":27698,"truncated":true},{"bytes":"# KABR Behavior Telemetry Data Dictionary\n\nThis document provides detailed descriptions of all data files and their fields in the KABR Behavior Telemetry dataset.\n\n## Dataset Files\n\n### Core Data Files\n\n#### `data/occurrences/`\nFrame-by-frame occurrence records for each video, combining detection tracks, behavior annotations, and telemetry data.\n\n**Files:** One CSV per video, named `{date}-{video_id}.csv` (e.g., `11_01_23-DJI_0977.csv`)\n\n**Key Fields:**\n- `date`: Recording date in DD_MM_YY format\n- `video_id`: DJI video identifier (e.g., DJI_0977)\n- `frame`: Frame number in the video sequence (0-indexed)\n- `date_time`: Timestamp in format \"YYYY-MM-DD HH:MM:SS,milliseconds,microseconds\"\n- `id`: Mini-scene identifier (track ID for behavioral sequences)\n- `latitude`: GPS latitude in decimal degrees (WGS84)\n- `longitude`: GPS longitude in decimal degrees (WGS84)\n- `altitude`: Altitude in meters above sea level\n- `iso`: Camera ISO setting\n- `shutter`: Shutter speed value\n- `fnum`: Aperture f-number\n- `ev`: Exposure value\n- `ct`: Color temperature\n- `color_md`: Color mode descriptor\n- `focal_len`: Lens focal length in millimeters\n- `dzoom_ratio`: Digital zoom ratio\n- `xtl`, `ytl`, `xbr`, `ybr`: Bounding box coordinates (top-left and bottom-right)\n- `z_order_x`, `z_order_y`: Display depth ordering for overlapping annotations\n- `label`: Object class label (e.g., \"zebra\", \"giraffe\")\n- `source`: Annotation source\n- `keyframe_x`, `keyframe_y`: Whether frame is a tracking keyframe\n- `outside_x`, `outside_y`: Whether object is outside frame bounds\n- `occluded_x`, `occluded_y`: Whether object is occluded\n- `points`: Polygon points for behavioral sequences\n- `behaviour`: Behavioral classification (e.g., \"walking\", \"grazing\", \"running\")\n\n**Record Count:** Varies by video; typically 10,000-66,000 frames per video\n**Coverage:** 47 videos with complete data\n**Missing Values:** Frames without detections/annotations have null values for annotation fields\n\n---\n\n#### `data/video_events.csv`\nDarwin Core Event records for individual video recordings.\n\n**Record Count:** 68 videos (47 with occurrence data)\n\n**Fields:**\n\n**Identifiers:**\n- `eventID`: Unique event identifier (format: \"KABR-2023:{session}:{video_id}\")\n- `parentEventID`: Links to session_events (format: \"KABR-2023:{session}\")\n\n**Event Classification:**\n- `eventType`: Type of sampling event (\"video recording\")\n- `eventDate`: Recording date (ISO 8601: YYYY-MM-DD)\n- `eventTime`: Video start time (HH:MM:SS)\n- `endTime`: Video end time (HH:MM:SS)\n\n**Geographic Information:**\n- `decimalLatitude`: Launch point latitude in decimal degrees (WGS84)\n- `decimalLongitude`: Launch point longitude in decimal degrees (WGS84)\n- `geodeticDatum`: Coordinate reference system (\"WGS84\")\n- `minimumElevationInMeters`: Minimum altitude during video\n- `maximumElevationInMeters`: Maximum altitude during video\n- `footprintWKT`: Geographic bounding box in Well-Known Text format\n\n**Sampling Protocol:**\n- `samplingProtocol`: \"Continuous aerial video recording\"\n\n**Associated Resources:**\n- `associatedMedia`: JSON object with paths to detection and behavior annotation files\n  ```json\n  {\n    \"detection\": \"path/to/DJI_XXXX_tracks.xml\",\n    \"behavior\": [\"path/to/trackID.xml\", ...]\n  }\n  ```\n\n**Remarks:**\n- `eventRemarks`: Description of video file (e.g., \"Video file DJI_0977.MP4\")\n\n---\n\n#### `data/session_events.csv`\nDarwin Core Event records for field sessions (missions/flights).\n\n**Record Count:** 17 sessions (14 with occurrence data)\n\n**Fields:**\n\n**Identifiers and Classification:**\n- `eventID`: Unique session identifier (format: \"KABR-2023:{date}_{session}\")\n- `parentEventID`: Dataset identifier (\"KABR-2023\")\n- `eventType`: Type of event (\"drone survey\")\n\n**Temporal Coverage:**\n- `eventDate`: Session date (ISO 8601: YYYY-MM-DD)\n- `eventTime`: Session start time (HH:MM:SS)\n- `endTime`: Session end time (HH:MM:SS)\n- `year`, `month`, `day`: Parsed date components\n\n**Geographic Coverage:**\n- `launchLatitude`: Drone launch point latitude (from first video)\n- `launchLongitude`: Drone launch point longitude (from first vi","encoding":"utf8","key":"metadata/DATA_DICTIONARY.md","name":"DATA_DICTIONARY.md","size":12607,"truncated":true}],"sourceObjectKey":"data/consolidated_metadata.csv","starterCell":"import pandas as pd\ndf = pd.read_csv(\"hf://datasets/imageomics/kabr-behavior-telemetry/data/consolidated_metadata.csv\")\ndf.head()","starterSamplePath":"data/consolidated_metadata.csv","totalObjects":65},"summary":"Frame-synchronised drone telemetry, animal detection boxes and expert behaviour annotations from aerial monitoring of Grevy's zebras (Equus grevyi), plains zebras (Equus quagga)…","title":"KABR Behavior Telemetry: Frame-Level Drone Telemetry, Detections and Behaviour Labels (Darwin Core)","updatedAt":"2026-11-16T16:40:26.944Z","visibility":"public","websiteUrl":"https://imageomics.github.io/KABR/"}}