StyleCast: an AI outfit planner that runs on your phone

Description and use

StyleCast automatically tags wardrobe items, reads local weather, remembers wear history, spots gaps, and suggests outfits or budget‑constrained purchases, all on the device.

Problem

Users manually guess outfits based on weather, leading to mismatches and wasted time.

Outcome

StyleCast improved outfit accuracy from 55–72% to 77% for clothing type and 90–93% for category, reduced model download to ~300 MB, and enabled fully local, privacy‑preserving outfit planning and budget shopping.

Workflow

• Snap wardrobe once; browser runs CLIP ViT‑B/32 to tag items.
• Read local weather via Open‑Meteo and user’s planned occasion.
• Apply JavaScript rules to pick an outfit, avoiding recent repeats and items in the wash.
• Detect gaps in wardrobe for current weather.
• If user asks for budget shopping, Gemma 3 runs locally to plan searches, which a Render server executes via SerpApi.
• Display outfit or shopping suggestions to the user.

Evidence

Snap your wardrobe once. An open vision model running in the browser tags each item: type, colour, warmth, waterproof, formality.

Source

CLIP doesn’t describe photos the way a chat model does. It compares a photo with a list of descriptions and picks the closest.

Source

Ask "I have €50, what should I buy?" and this happens: Gemma reads your wardrobe and the gaps the rules found, and plans up to three searches.

Source