suryadv/BirdSnap_masked

BirdSnap Masked

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

File

File	Description
masks.parquet	37,693 rows — one per accepted mask — with RLE mask, score, source image identifier, normalized bird species/common-name label, raw BirdSnap label, split, and preserved source row metadata

Schema

Column	Type	Description
dataset	str	Always BirdSnap
text_prompt	str	Text prompt used to generate the mask, always bird
mask_rle_counts	str	RLE-encoded mask counts in 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
image_id	str	BirdSnap source stream identifier, e.g. BirdSnap_000000
file_name	str	Source file name when exposed by the Hugging Face row, otherwise the same synthetic BirdSnap identifier as image_id
species	str	Bird species/common name normalized from the raw label, with underscores replaced by spaces
common_name	str	Bird species/common name normalized from the raw label, with underscores replaced by spaces
label	str	Raw BirdSnap label from the Hugging Face dataset, e.g. Greater_White_fronted_Goose
split	str	Source Hugging Face split, always train
original_metadata	str	JSON-encoded non-image metadata preserved from the source Hugging Face row

Loading the dataset

Load the Parquet file directly from Hugging Face.

import pandas as pd

df = pd.read_parquet(
    "hf://datasets/suryadv/BirdSnap_masked/masks.parquet"
)

print(len(df))
print(df.columns.tolist())

Retrieving the original image

The image_id column stores the source row index in the sasha/birdsnap Hugging Face train stream.

from itertools import islice

from datasets import load_dataset

row = df.iloc[0]

source_index = int(row["image_id"].rsplit("_", 1)[1])

source = load_dataset(
    "sasha/birdsnap",
    split="train",
    streaming=True,
)

sample = next(islice(source, source_index, None))
img = sample["image"].convert("RGB")

img

Retrieving an image by species

Look up an example image corresponding to a particular BirdSnap species/common name.

from itertools import islice

from datasets import load_dataset

species = "Wilsons Phalarope"

row = df[
    df["species"].str.lower() == species.lower()
].iloc[0]

source_index = int(row["image_id"].rsplit("_", 1)[1])

source = load_dataset(
    "sasha/birdsnap",
    split="train",
    streaming=True,
)

sample = next(islice(source, source_index, None))
img = sample["image"].convert("RGB")

img

Decoding a mask

Decode a COCO-format RLE mask into a binary NumPy array.

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)

print(binary_mask.shape)

The decoded mask is a NumPy array with shape (H, W) and dtype uint8.

License

Please consult the original BirdSnap / Hugging Face dataset terms when using the source imagery or metadata.

If you use BirdSnap, please also cite the original BirdSnap CVPR 2014 paper.