metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: instances
      list:
        - name: box
          sequence: float64
        - name: label
          dtype: int64
  splits:
    - name: train
      num_bytes: 25165898.049
      num_examples: 1107
  download_size: 0
  dataset_size: 25165898.049
license: apache-2.0
task_categories:
  - object-detection
language:
  - en
pretty_name: wheel_chair_detection
size_categories:
  - 1K<n<10K

Wheelchair Dataset for Object Detection

Dataset Information

The dataset_info file provides information about the wheelchair dataset designed for object detection. Here are the details:

Features

image: Represents the images in the dataset.
- Data type: image
instances: Represents the instances within each image. Each instance consists of a bounding box and a label.
- Data type: list
- Sub-features:
  - box: Bounding box coordinates for each instance.
    - Data type: float64
  - label: Label for each instance.
    - Data type: int64

Splits

Train: This split, named "train," contains a total of 1,107 examples.
- Number of bytes: 25,165,898.049
- Number of examples: 1,107

Dataset Size

Download size: 0 (no download required)
Dataset size: 25,165,898.049 bytes

Wheelchair Class Name

The dataset includes the following class names for object detection:

"labels": ClassLabel(names=["person", "wheel_chair", "not wheel chair"])

The class labels are defined as follows:

"person"
"wheel_chair"
"not wheel chair"

Object Detection Application (YOLOv Models)

You can utilize the dataset with YOLOv models for object detection tasks. The class labels for the models correspond to the defined class names mentioned above:

"labels": ClassLabel(names=["person", "wheel_chair", "not wheel chair"])

Make sure to follow the appropriate implementation guidelines and examples for YOLOv models to leverage this dataset effectively.

loading the Dataset

'''

Load the dataset

hf_dataset = load_dataset("Falah/wheel-chair-images-annotation4object-detec", split="train")

Accessing image

image = hf_dataset[1]['image']

Display the image

image.show()

Accessing label and bounding box coordinates

instances = hf_dataset[1]['instances'] for instance in instances: label = instance['label'] box = instance['box']

# Get the class name for the label
class_name = hf_dataset.features['instances']['label'].int2str(label)
print(f"Label: {class_name}")
print(f"Bounding Box: {box}")

'''

Dataset Citation

If you use this wheelchair dataset for object detection in your work, please cite it using the following template:

@dataset{falah_salieh_wheelchair_dataset_2023,
  author = {Falah G. Salieh},
  title = {Wheelchair Dataset for Object Detection},
  year = {2023},
  publisher = {Huggingface.co},
  url = {\url{https://huggingface.co/datasets/Falah/wheel-chair-images-annotation4object-detec}},
}