Genius-Society/tt100k

Intro

The Tsinghua–Tencent 100K (TT100K) dataset is a large-scale traffic sign benchmark built from 100,000 Tencent Street View images, containing over 30,000 annotated instances across 221 categories. Designed for real-world detection and classification tasks, it features significant variation in lighting, weather, viewpoint, and distance. While the original study focuses on a 45-class subset using a 100-instance threshold per class, the Ultralytics configuration retains all 221 categories, including many with sparse samples, making the dataset both comprehensive and challenging for robust model development.

Usage

from datasets import load_dataset
ds = load_dataset(
    "Genius-Society/tt100k",
    name="default",
    split="train",
    cache_dir="./__pycache__",
)
for i in ds:
    print(i)

Maintenance

GIT_LFS_SKIP_SMUDGE=1 git clone git@hf.co:datasets/Genius-Society/tt100k
cd tt100k

Mirror

https://www.modelscope.cn/datasets/Genius-Society/tt100k

Thanks

• https://docs.ultralytics.com/datasets/detect/tt100k