Mumind93/jewelry-design-dataset

💎 Jewelry Text-to-Image Fine-Tuning Dataset

This dataset is created for fine-tuning Stable Diffusion XL (SDXL) models to generate realistic and diverse images of jewelry — including rings, bracelets, necklaces, and earrings.

📦 Dataset Summary

The dataset includes over 6,100 high-resolution images of jewelry, organized into four categories. It was collected from various sources and manually curated. Each image is associated with a simple or structured caption describing the type of jewelry for text-to-image fine-tuning.

📁 Structure

jewelry-sdxl-dataset/:
  - bracelet/               # 888 images
  - earring_best/           # 3,298 images
  - necklace/               # 1,738 images
  - ring_best/              # 233 images
  - captions.csv or .txt    # (Optional) image-to-text mapping for training

You can convert folder names into prompts like:

• —a high quality photo of a gold bracelet
• —a stylish diamond ring on white background

🧠 Intended Uses

• —Fine-tuning Stable Diffusion XL for jewelry generation
• —Fashion product mockups & concept design
• —E-commerce image augmentation
• —AI-generated advertising material
• —Generative research in domain-specific design

🧪 Example Prompts

You can use this dataset with --train_text_encoder and prompts like:

• —a luxury diamond ring on a black velvet
• —a pair of gold earrings with pearls
• —a necklace with a heart-shaped pendant

🔓 License

This dataset is available under the MIT License. You are free to use it in research, commercial, and open-source projects with attribution.

📬 Creator

Siddharth Patel (@sidd707) Built for generative AI research and diffusion model fine-tuning. Contact: sidd707888@gmail.com