Dataset Card for LogoDet-3K

LogoDet-3K dataset aims on logotype (image) detection task.

Dataset Description

LogoDet-3K consists of thousand images with brands' logotypes and their bounding boxes. This dataset aims to help train logotype detection models.

• License: MIT

Dataset Usage

You can download this dataset by the following command (make sure that you have installed Huggingface Datasets):

from datasets import load_dataset

dataset = load_dataset("PodYapolsky/LogoDet-3K")

Company ids mapping to names and vice versa

company2id = {
    name: idx
    for idx, name in enumerate(dataset["train"].features["company_name"].names)
}
id2company = {v: k for k, v in company2id.items()}

Dataset Structure

The dataset is provided in Parquet format and contains the following attributes:

{
    "image_path": [PIL.Image],
    "industy_name": [str] Industry type company's brand belongs to,
    "company_name": [int] The company id to which brand belongs,
    "bbox": [tuple[int]] bounding box in format ('xmin', 'ymin', 'xmax', 'ymax'),
}