File size: 3,956 Bytes
13c3710 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 |
# TestNew.py
import json
import random
from datasets import (
BuilderConfig,
DatasetInfo,
DownloadManager,
GeneratorBasedBuilder,
SplitGenerator,
Split,
Features,
Image,
Sequence,
Value,
)
from huggingface_hub import hf_hub_url
_REPO_ID = "iamirulofficial/TestNew" # change if you ever fork the repo
class ImageSubsetConfig(BuilderConfig):
"""BuilderConfig for full dataset vs. small random sample."""
def __init__(self, name, sample_size=None, **kwargs):
super().__init__(
name=name,
version="1.0.1",
description=kwargs.get("description", "")
)
self.sample_size = sample_size
class MyImageDataset(GeneratorBasedBuilder):
"""Images + 2‑D actuated_angle labels stored in metadata.json."""
BUILDER_CONFIGS = [
ImageSubsetConfig(
name="full",
sample_size=None, # all images
description="Entire dataset (≈100 GB)"
),
ImageSubsetConfig(
name="small",
sample_size=2, # tiny sample for quick tests
description="Two random images"
),
]
DEFAULT_CONFIG_NAME = "small"
def _info(self) -> DatasetInfo:
return DatasetInfo(
description="Images with a 2‑D actuated_angle from metadata.json",
features=Features(
{
"image": Image(), # PIL.Image will be returned
"actuated_angle": Sequence(Value("int32")), # [angle0, angle1]
}
),
supervised_keys=None,
)
# --------------------------------------------------------------------- #
# Download phase #
# --------------------------------------------------------------------- #
def _split_generators(self, dl_manager: DownloadManager):
# 1️⃣ Download metadata.json (tiny text file)
meta_path = dl_manager.download(
hf_hub_url(_REPO_ID, "metadata.json", repo_type="dataset")
)
# 2️⃣ Decide which filenames we need
with open(meta_path, encoding="utf-8") as f:
metadata = json.load(f) # {"frame_000.png": {"0":…, …}, …}
all_fnames = list(metadata)
if self.config.sample_size: # small‑config branch
random.seed(42)
selected = sorted(random.sample(all_fnames, self.config.sample_size))
else:
selected = sorted(all_fnames) # full dataset
# 3️⃣ Build URLs → dl_manager.download() → local paths
url_dict = {
fname: hf_hub_url(
_REPO_ID, f"images/{fname}", repo_type="dataset"
)
for fname in selected
}
img_paths = dl_manager.download(url_dict) # same keys, but local files
return [
SplitGenerator(
name=Split.TRAIN,
gen_kwargs={
"img_paths": img_paths,
"metadata": metadata,
},
)
]
# --------------------------------------------------------------------- #
# Generate examples #
# --------------------------------------------------------------------- #
def _generate_examples(self, img_paths: dict, metadata: dict):
"""
Yields (key, example) where example =
{ "image": <local‑file‑path>, "actuated_angle": [int, int] }
"""
for idx, (fname, local_path) in enumerate(img_paths.items()):
meta = metadata.get(fname, {})
angles = [int(meta.get("0", 0)), int(meta.get("1", 0))]
yield idx, {"image": local_path, "actuated_angle": angles}
|