AI Cashier
A self-checkout till for shops without an IT department. A camera recognises what the customer puts down, and a new product takes five photos and a price to teach.
- Type
- App
- Status
- In progress
- Year
- 2026
AI Cashier is a six-person school engineering project. The studio built all of its software: the till, the research tools, the website and the iPhone app. It won 3rd place at the KMUTT Science & Technology Idea Contest #4.
What it does
- A camera looks at the counter and matches each item against products the till has been taught. A weight check confirms the basket.
- No retraining and no model to manage. Owners teach a product with five photos and a price.
- Payment by PromptPay QR. Runs on Windows with any webcam, with a matching iPhone app.
How we built it
- Detect, embed and match pipeline in Python with ONNX models, a Qt till and SQLite.
- Our measurements: switching the image encoder to MobileCLIP-B raised top-1 accuracy on packaged goods from 45.8% to 89.6% with no fine-tuning.
- Windows installer built by GitHub Actions. iPhone app in SwiftUI with Core ML.
Screens
Have an app or website in mind?
Tell us what you want to build. We reply within 24 hours with questions or a plan.