BanglaOrna: A Smartphone Image Dataset of Traditional Bangladeshi Ornaments and Accessories

Traditional Bangladeshi ornaments and accessories—such as churi (terracotta bangles), tikli (filigree forehead ornaments), jhumka (bell-shaped earrings), and Komor Bicha (intricate waist chains)—are emblematic of the nation’s cultural heritage. Despite their anthropological and artistic value, no standardized dataset exists for computational research, hindering applications in heritage preservation, fashion technology, and ethnography.
We present BanglaOrna, the first comprehensive dataset with a total of 7,174 high-resolution smartphone images documenting 27 categories of women’s traditional adornments, including kaaner duul (dangling earrings), haar (layered necklaces), and nakphul (bridal nose pins). Collected from Dhaka’s historic markets (New Market, Gausia Market) under diverse lighting and angles, the dataset reflects real-world variability in ornament styles and materials.
Images were captured using Redmi Note 14 Pro (200 MP), Redmi Note 14 (108 MP), and Redmi Note 12 Pro+ (200 MP), ensuring fine-grained detail for small, intricate designs. Each image is annotated in COCO format with bounding boxes. Augmentation techniques (Horizontal and Vertical flip (50%), (±10 °) Horizontal and Vertical Shear,Brightness (±25%), Noise ( 0.5% )) were applied to improve model robustness.
BanglaOrna opens doors to: 
Training AI models for automated ornament classification and heritage documentation,
Supporting ethnographic research on regional design variations and 
Applications in virtual try-ons for e-commerce and digital museum archives.