Horses or Humans Dataset
Dataset composed of 1,283 photorealistic images representing horses and humans in various poses and environments. Includes ethnic and gender diversity for humans. Used to train binary classifiers in computer vision.
1,283 300x300 pixel images in JPEG/PNG format
Creative Commons Attribution 2.0 (CC BY 2.0)
Description
The dataset Horses or Humans contains 300x300 pixel images divided into two classes: horses and humans. It is designed to train deep neural networks to distinguish between these two categories. The dataset focuses on the diversity of poses, environments and human characteristics (gender, ethnicity).
What is this dataset for?
- Training binary classifiers in computer vision
- Test the ability to generalize models on photorealistic images
- Deep Learning and CNN Educational Support
Can it be enriched or improved?
Yes, the dataset can be completed with annotations on poses, angles of view, or segmentation for more advanced tasks. Increasing data can also improve the robustness of models.
🔎 In summary
🧠 Recommended for
- AI students
- CV researchers
- Deep learning application developers
🔧 Compatible tools
- TensorFlow
- PyTorch
- Keras
- OpenCV
💡 Tip
Use image augmentation techniques to improve the robustness of models.
Frequently Asked Questions
What is the format and resolution of the images in this dataset?
The images are in JPEG/PNG format, all 300x300 pixels.
Is this dataset suitable for supervised learning?
Yes, it is designed to train binary supervised classifiers.
Can this dataset be used for tasks other than binary classification?
Yes, but it will require additional annotations, for example for segmentation or pose recognition.
