F-Wireframe Dataset

✏️ Github   |   📑 Paper    |   🖼️ Viewer

This is the F-Wireframe dataset, designed for fisheye image line segment detection.

Summary

The F-Wireframe dataset is derived from the Wireframe dataset by distorting images with the fisheye distortion model, where the fisheye distortion coefficient of each image is randomly generated.

Number of samples:

• Train: 5,000
• Test: 462

Download

• Download with huggingface-hub

python3 -m pip install huggingface-hub
huggingface-cli download --repo-type dataset lh9171338/F-Wireframe --local-dir ./

• Download with Git

git lfs install
git clone https://huggingface.co/datasets/lh9171338/F-Wireframe

Usage

• Load the dataset from Hugging Face Hub

from datasets import load_dataset

ds = load_dataset("lh9171338/F-Wireframe")
# or load from `refs/convert/parquet` for acceleration
# from datasets import load_dataset, Features, Image, Sequence, Value
# features = Features({
#     "image": Image(),
#     "image_file": Value("string"),
#     "image_size": Sequence(Value("int32")),
#     "camera_type": Value("string"),
#     "camera_coeff": {
#         "K": Sequence(Sequence(Value("float32"))),
#         "D": Sequence(Value("float32")),
#     },
#     "lines": Sequence(Sequence(Sequence(Value("float32")))),
# })
# ds = load_dataset("lh9171338/F-Wireframe", features=features, revision="refs/convert/parquet")
print(ds)
# DatasetDict({
#     train: Dataset({
#         features: ['image', 'image_file', 'image_size', 'camera_type', 'camera_coeff', 'lines'],
#         num_rows: 5000
#     })
#     test: Dataset({
#         features: ['image', 'image_file', 'image_size', 'camera_type', 'camera_coeff', 'lines'],
#         num_rows: 462
#     })
# })
print(ds["test"][0].keys())
# dict_keys(['file_name', 'image_file', 'image_size', 'camera_type', 'camera_coeff', 'lines'])

• Load the dataset from local

from datasets import load_dataset

ds = load_dataset("imagefolder", data_dir=".")
print(ds)
# DatasetDict({
#     train: Dataset({
#         features: ['image', 'image_file', 'image_size', 'camera_type', 'camera_coeff', 'lines'],
#         num_rows: 5000
#     })
#     test: Dataset({
#         features: ['image', 'image_file', 'image_size', 'camera_type', 'camera_coeff', 'lines'],
#         num_rows: 462
#     })
# })
print(ds["test"][0].keys())
# dict_keys(['file_name', 'image_file', 'image_size', 'camera_type', 'camera_coeff', 'lines'])

• Load the dataset with jsonl files

import jsonlines

with jsonlines.open("test/metadata.jsonl") as reader:
    infos = list(reader)
print(infos[0].keys())
# dict_keys(['file_name', 'image_file', 'image_size', 'camera_type', 'camera_coeff', 'lines'])

Citation

@article{LI2021187,
    title = {ULSD: Unified line segment detection across pinhole, fisheye, and spherical cameras},
    author = {Hao Li and Huai Yu and Jinwang Wang and Wen Yang and Lei Yu and Sebastian Scherer},
    journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
    volume = {178},
    pages = {187-202},
    year = {2021},
}