This model classifies furniture and home product images into a set of predefined style categories using a confidence-aware classification approach.

Supported style categories include: - Glam / Luxury - Minimal / Scandi - Industrial / Loft - Retro / Mid-Century - Romantic / Vintage

Instead of forcing every image into a single style, the model evaluates prediction confidence and may return multiple styles or no label when the input does not strongly match any known category. This helps reduce incorrect or misleading classifications for ambiguous, low-quality, or out-of-domain images.

The model is designed for real-world furniture catalogs and home product pipelines, where reliability is critical. It is suitable for applications such as catalog enrichment, search filtering, style-based browsing, and recommendation systems.