Playing Cards Classification Dataset
High-resolution image dataset of playing cards, including 53 categories (each unique card) with a total of 8,157 clean and well-centered JPEG images.
Description
The Playing Cards Classification Dataset contains over 8,000 playing card images, with each image representing a unique card (including the joker). The files are classified into 53 folders corresponding to the 53 types of cards, with a clear division between training, validation, and test sets. All images are 224×224 in size, ready for classification models.
What is this dataset for?
- Train image classification models (CNN, MobileNet...)
- Create card recognition systems for automated games or fun applications
- Computer vision educational projects for beginners
Can it be enriched or improved?
Yes, the dataset can be extended with various angles of view, different lighting conditions, or more realistic backgrounds. It is also possible to add additional annotations such as color or position on a table for more complex projects.
🔎 In summary
🧠 Recommended for
- AI students
- Game app developers
- Computer vision training
🔧 Compatible tools
- TensorFlow
- PyTorch
- Keras
- FastAI
- Roboflow
💡 Tip
Use the subfolder structure for automatic loading with ImageDataGenerator or Datasets.ImageFolder.
Frequently Asked Questions
Does the dataset contain all the cards in a complete game?
Yes, all 52 classic cards plus one joker are represented, for a total of 53 classes.
Are the images all the same size and size?
Yes, they are all in JPG format, in 224×224 pixels and in 3 color channels.
Can it be used for real-time recognition?
Yes, with a lightweight model (e.g. MobileNet), this dataset can be used to train a real-time classifier on camera or mobile.
