7000 Dress Style Pictures
Dataset of images of women's clothing with 7240 images in 640x640 format, including random increases (rotation, blur, noise, etc.). Ideal for e-commerce or visual AI projects.
Description
This dataset contains 7240 images of feminine dresses, pretreated to have a uniform size (640x640). Each frame was subjected to various increases: random rotation, Gaussian blur, salt and pepper noise, or horizontal flip. This visual corpus is particularly suitable for computer vision projects in the field of fashion or online commerce.
What is this dataset for?
- Training object recognition models for fashion
- Building a style or outfit recommendation engine
- Create prototypes for virtual try-on or look analysis
Can it be enriched or improved?
Yes. It is possible to add labels (category, color, type of dress), or to enrich with metadata from a product catalog. Annotations (bounding boxes, masks) could also transform it into a dataset useful for detection or segmentation.
🔎 In summary
🧠 Recommended for
- Data scientists in e-commerce
- Computer vision researchers
- Student projects
🔧 Compatible tools
- PyTorch
- TensorFlow
- Keras
- FastAI
💡 Tip
To improve performance, add your own labels or combine with an annotated dataset.
Frequently Asked Questions
Does this dataset contain clothing categories?
No, the images are not labelled. However, you can add your own classes via manual annotation or a semi-supervised model.
Can I use it for object detection or segmentation?
Not directly, but you can manually annotate images or combine them with a segmented dataset to train a model.
Are the images adapted to pre-trained models like ResNet or MobileNet?
Yes, the 640x640 resolution is compatible with many CNN models. No additional resizing is required.
