README.md

metadata

dataset_info:
  features:
    - name: file_name
      dtype: string
    - name: image
      dtype: image
    - name: id
      dtype: int64
    - name: category_id
      dtype:
        class_label:
          names:
            '0': Tumor
            '1': '0'
            '2': '1'
    - name: bbox
      sequence: float32
      length: 4
    - name: segmentation
      sequence:
        sequence: float32
    - name: area
      dtype: float32
    - name: iscrowd
      dtype: int64
    - name: height
      dtype: int64
    - name: width
      dtype: int64
    - name: date_captured
      dtype: string
    - name: license
      dtype: int64
  splits:
    - name: train
      num_bytes: 113682589.25
      num_examples: 1502
    - name: test
      num_bytes: 16317010
      num_examples: 215
    - name: valid
      num_bytes: 32320851
      num_examples: 429
  download_size: 161781456
  dataset_size: 162320450.25

Dataset Card for "brain-tumor-image-dataset-semantic-segmentation"

Dataset Description

The Brain Tumor Image Dataset (BTID) for Semantic Segmentation contains MRI images and annotations aimed at training and evaluating segmentation models. This dataset was sourced from Kaggle and includes detailed segmentation masks indicating the presence and boundaries of brain tumors.

This dataset can be used for developing and benchmarking algorithms for medical image segmentation, particularly in identifying and segmenting brain tumors.

Features

Field Name	Data Type	Description	Example Value	Usage
file_name	String	Name of the image file	"2256_jpg.rf.3afd7903eaf3f3c5aa8da4bbb928bc19.jpg"	Reference to the image file
image	Image	Image data	<PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=640x640>	Input for image analysis
id	Integer	Unique identifier for each image	0	Image-level identification
category_id	ClassLabel	Class label indicating if the image is of a tumor or not	1 (where 0 = 'Tumor', 1 = 'Normal')	Classification of the image
bbox	List[Float]	Bounding box coordinates for the tumor	[145.0, 239.0, 168.75, 162.5]	Object detection, region of interest
segmentation	List[List[Float]]	Segmentation mask coordinates for the tumor	[[313.75, 238.75, 145.0, 238.75, 145.0, 401.25, 313.75, 401.25, 313.75, 238.75]]	Detailed object segmentation
area	Float	Area covered by the tumor	27421.875	Feature for analysis
iscrowd	Integer	Indicates if the segmentation mask is for a crowd	0	Binary flag for single object/crowd
height	Integer	Height of the image	640	Image dimension
width	Integer	Width of the image	640	Image dimension
date_captured	String	Date when the image was captured	"2023-08-19T04:37:54+00:00"	Metadata for the image
license	Integer	License information for the image	1	License reference for usage rights

This table provides a clear and concise description of each field, its data type, an example value, and its usage within the dataset.

Data Splits

The dataset is split into three subsets:

• Train: 1502 images
• Validation: 429 images
• Test: 215 images

Example Usage

Here is how you can load and visualize the dataset using the Hugging Face datasets library:

from datasets import load_dataset

dataset = load_dataset("dwb2023/brain-tumor-image-dataset-semantic-segmentation")
image = dataset["train"][0]["image"]
image.show()

Citation

Please cite the original creator of the dataset:

@dataset{Darabi2023BrainTumor,
  author       = {Peyman Darabi},
  title        = {Brain Tumor Image Dataset : Semantic Segmentation},
  month        = aug,
  year         = 2023,
  url          = {https://www.kaggle.com/datasets/pkdarabi/brain-tumor-image-dataset-semantic-segmentation},
  note         = {The Tumor Segmentation Dataset is designed specifically for the TumorSeg Computer Vision Project, which focuses on Semantic Segmentation. The project aims to identify tumor regions accurately within Medical Images using advanced techniques. The dataset contains two classes: Tumor (Class 1) and Non-Tumor (Class 0).},
}

License

The dataset is licensed under CC BY-NC 4.0. For more information, visit this link.