kaityc06/TreeOfLife-200M_masked

TreeOfLife-200M Masked

Segmentation masks for the TreeOfLife-200M dataset, stored as RLE-encoded masks in a single Parquet file.

File

File	Description
masks.parquet	133,850 rows — one per image — with RLE mask, score, taxonomic labels, and uuid

Schema

Column	Type	Description
uuid	str	UUID identifying the image (matches the TreeOfLife-200M record)
dataset	str	Always TreeOfLife-200M
text_prompt	str	Text prompt used to generate the mask
mask_rle_counts	str	RLE-encoded mask (COCO format)
mask_rle_height	int	Height used for RLE decoding
mask_rle_width	int	Width used for RLE decoding
mask_score	float	Mask confidence score
image_width	int	Original image width
image_height	int	Original image height
source_url	str	Direct URL to the original image
kingdom	str	Taxonomic kingdom (e.g. Animalia, Plantae, Fungi)
phylum	str	Taxonomic phylum
class	str	Taxonomic class
order	str	Taxonomic order
family	str	Taxonomic family
genus	str	Taxonomic genus
species	str	Species epithet
scientific_name	str	Full scientific name (genus + species)
common	str	Common name
data_source	str	Source database (gbif or eol)
publisher	str	Data publisher (e.g. iNaturalist.org)
basis_of_record	str	How the record was collected (e.g. HUMAN_OBSERVATION)
img_type	str	Image provenance (Citizen Science or Museum Specimen: ...)
source_id	str	Record ID in the source database
shard_filename	str	Tar shard file containing the image
shard_file_path	str	Relative path to the shard within the dataset
base_dataset_file_path	str	Path to the HDF5 file in the base dataset

Loading the masks

import pandas as pd

df = pd.read_parquet("hf://datasets/kaityc06/TreeOfLife-200M_masked/masks.parquet")

Retrieving the original image from a uuid

Each row contains a source_url pointing directly to the original image, so no lookup table is needed:

import requests
from PIL import Image
from io import BytesIO

row = df.iloc[0]
response = requests.get(row["source_url"])
img = Image.open(BytesIO(response.content))

Decoding a mask

import numpy as np
from pycocotools import mask as mask_utils

row = df.iloc[0]

rle = {
    "counts": row["mask_rle_counts"],
    "size": [row["mask_rle_height"], row["mask_rle_width"]],
}
binary_mask = mask_utils.decode(rle)  # numpy array, shape (H, W), dtype uint8