KadaCard: WhatsApp listings from a product photo, scored offline with open-weight CLIP
Описание и применение
Kadacard позволяет мелким продавцам в Керале быстро создавать двуязычные листинги для WhatsApp, используя фотографию товара и локальную модель CLIP без подключения к серверу.
Задача
Kerala micro‑sellers need a zero‑cost, privacy‑preserving way to turn a product photo into a bilingual WhatsApp listing without a backend or third‑party API.
Результат
Kadacard enables micro‑sellers to generate accurate, bilingual WhatsApp listings from a single photo, with no backend, no data leakage, and zero per‑image cost.
Рабочий процесс
- Open Kadacard in the browser
- Take a photo of the product
- The on‑device CLIP model classifies the image into one of nine niches
- The app displays the top three niche suggestions and auto‑fills a bilingual template
- Seller enters price and WhatsApp number
- Tap copy to copy the listing text
- Paste into WhatsApp Business
Доказательства
KadaCard (കട means shop) is a small browser tool for Kerala micro‑sellers. You photograph what you sell — banana chips, pickle, spices, handloom, bakery — and an open-weight CLIP model running on your phone picks a niche.
There is no backend. Photos are scored in the tab and discarded. Only the listing text you choose to save stays in localStorage.
The category check is Xenova/clip-vit-base-patch32, an ONNX build of CLIP, loaded with Transformers.js. The weights are open, so the check does not depend on a proprietary vision API, an account, or a server that can see the product photo.
- Дата записи: 2026-10-13T05:40:56.132Z