ravisri/BirdBinaryClassification

---

dataset_info:
features:
- name: image
dtype: string
- name: image_id
dtype: string
- name: caption
dtype: string
- name: label
dtype: string
license: "VIT"

Bird Detection Dataset

This dataset contains images and corresponding labels for a binary classification task: identifying whether a given image contains a bird or not. It is designed for use in computer vision tasks such as image classification, detection, or transfer learning.

Dataset Structure

• Total Images: 4,410
  • 2,973 images containing birds (bird)
  • 1,437 images not containing birds (no_bird)
• Split: 80% training (train.csv), 20% testing (test.csv)

Folder Layout

my-bird-dataset/ ├── train.csv ├── test.csv ├── bird/ # Contains images named bird-0001.jpg, ..., bird-2973.jpg ├── no_bird/ # Contains images named no_bird-0001.jpg, ..., no_bird-1437.jpg

less

CSV Format

Each CSV (train.csv or test.csv) has the following columns:

image	image_id	caption	label
path/to/image.jpg	bird-0001	This image contains a bird.	bird
path/to/image.jpg	no_bird-0001	This image does not contain a bird.	no_bird

• image: Relative path to the image file
• image_id: Unique image identifier
• caption: Description of the image contents
• label: Class label (bird or no_bird)

How to Use

You can load this dataset using the datasets library:

from datasets import load_dataset  

dataset = load_dataset("ravisri/BirdBinaryClassification")  
train = dataset['train']  
test = dataset['test']  
Or, load the CSV directly:

python
train = load_dataset("ravisri/BirdBinaryClassification", data_files="train.csv")['train']  
License
Include your license here (e.g., MIT, CC BY 4.0). If unknown, please specify.

Citation
If you use this dataset in your work, please cite:

ini
@dataset{your_name_2024_bird_detection,  
  title = {BirdBinaryClassification},  
  author = {ravisri},  
  year = {2024},  
  url = {https://huggingface.co/datasets/ravisri/BirdBinaryClassification},  
}  
Contact
For questions, please open an issue or contact [ravisripallam@gmail.com].