Datasetpublic

gfermoto/birdlense-annotations

BirdLense Annotations Dataset of bird annotations from feeders for the BirdLense Hub project — a bird feeder monitoring system with detection (YOLO, BirdNET). Description Images of birds and squirrels captured by BirdLense cameras at feeders. Annotations in YOLO format (bounding boxes) for training and fine-tuning detection models. Data is collected by community members through confirming and correcting automatic detections. Dataset Structure train/… See the full description on the dataset page: https://huggingface.co/datasets/gfermoto/birdlense-annotations.

sourceHugging Facecc-by-nc-nd-4.0updated 6mo agoView on Hugging Face

0likes18downloads

Dataset Card

BirdLense Annotations

Dataset of bird annotations from feeders for the BirdLense Hub project — a bird feeder monitoring system with detection (YOLO, BirdNET).

Description

Images of birds and squirrels captured by BirdLense cameras at feeders. Annotations in YOLO format (bounding boxes) for training and fine-tuning detection models. Data is collected by community members through confirming and correcting automatic detections.

Dataset Structure

train/
  images/     # Training images
  labels/     # YOLO annotations (.txt)
val/
  images/
  labels/
data.yaml     # Class configuration and paths

Classes

Species list is defined in data.yaml. Typical categories: birds (tits, sparrows, woodpeckers, etc.), squirrels, mice.

Usage

Download

python

from huggingface_hub import snapshot_download

path = snapshot_download(
    repo_id="gfermoto/birdlense-annotations",
    repo_type="dataset",
    local_dir="./birdlense-data",
)

YOLO Training

bash

yolo detect train data=./birdlense-data/data.yaml model=yolov8n.pt epochs=100

Data Source

—BirdLense Hub — open-source bird feeder monitoring system
—Detection: YOLO + BirdNET
—Labeling: user confirmation/correction in the UI

License

CC BY-NC-ND 4.0 — attribution required, non-commercial, no derivatives without permission.