CoolFace
Datasetpublic

the-shoaib2/geotree

Tree Monitoring Dataset - Bangladesh This dataset is prepared for training object detection and segmentation models to monitor tree canopies in Bangladesh. Dataset Structure sentinel2/: Raw Sentinel-2 Level-2A imagery for Bandarban, Rangamati, Sylhet, and Gazipur districts. deepforest/: DeepForest annotations and tree crown samples. zenodo/: Reference training datasets. selvabox/: Tree canopy labels and annotations. global_forest_change/: Hansen Global Forest… See the full description on the dataset page: https://huggingface.co/datasets/the-shoaib2/geotree.

sourceHugging Facecc-by-4.0updated 2mo agoView on Hugging Face
0likes103downloads
Dataset Card

Tree Monitoring Dataset - Bangladesh

This dataset is prepared for training object detection and segmentation models to monitor tree canopies in Bangladesh.

Dataset Structure

  • —sentinel2/: Raw Sentinel-2 Level-2A imagery for Bandarban, Rangamati, Sylhet, and Gazipur districts.
  • —deepforest/: DeepForest annotations and tree crown samples.
  • —zenodo/: Reference training datasets.
  • —selvabox/: Tree canopy labels and annotations.
  • —global_forest_change/: Hansen Global Forest Change tiles covering Bangladesh.
  • —yolo/: Formatted images and labels ready for training YOLO.
  • —coco/: Formatted images and COCO JSON annotations.
  • —segmentation/: Formatted images and binary masks for semantic segmentation.
  • —labels/: Raw annotations and vector label formats.

Classes

  1. 1.tree - Canopy crown boundary

How to use with YOLOv8

python
from ultralytics import YOLO
model = YOLO('yolov8n.pt')
model.train(data='dataset.yaml', epochs=100, imgsz=640)