AI Multi-Item Recognition
with KM-1x16 Kiosk
Recognize multiple products in a single image and turn a full tray into an accurate checkout list within seconds.
Combining a high-resolution smart camera, dedicated AI processing and a trainable deep-learning algorithm, the system helps retailers reduce manual input and deliver a smoother checkout experience.
PowerfulAI Chip
SmartCamera
InstantRecognition
Fast TotalCalculation
Just Place the Tray.
AI Handles the Rest.
Instead of scanning products individually, customers or staff simply place the items within the recognition area. The system captures the entire tray, identifies multiple products simultaneously, and sends the results to the checkout interface for confirmation and total calculation.
It is an intuitive, contact-free workflow designed for bakeries, donut shops, self-serve restaurants and other retail environments where speed and convenience matter.
Multiple Items.
One Intelligent Capture.
Powered by a trainable YOLO-based deep-learning algorithm, the system detects and classifies multiple products within a single image. Each recognized item can be matched with its corresponding product record, price or PLU information.
The results are then displayed on screen, allowing staff or customers to review the order and complete payment without entering every item manually.
Simultaneous Multi-item Recognition
Identify different products across the tray in a single scan, eliminating the need for one-by-one barcode scanning.
Trainable Product Models
Build recognition models around your own product catalog. The system can be updated as new products are introduced or existing products change in appearance.
Confidence-based Detection
Each detected item is evaluated by the AI model, helping the checkout application preserve clear and reviewable recognition results.
Continuous Software Improvement
Recognition models and product data can be refined through software updates, enabling the solution to evolve alongside changing menus and retail assortments.
AI-Powered Vision for Reliable Recognition
The system combines the RV1126B AI vision processor with a high-resolution 5-megapixel camera to deliver fast and reliable multi-item recognition.
Its quad-core ARM Cortex-A53 CPU and dedicated 3 TOPS NPU accelerate deep-learning inference, while the 2592 × 1944 image sensor captures detailed tray images at up to 30 frames per second.
A 4G+IR lens, automatic exposure, gain control and white balance help maintain consistent image quality across everyday checkout conditions.
With a working distance of approximately 60 cm and a focused field of view, the camera is optimized for top-down product recognition in retail environments.
AI and Imaging Highlights
- RV1126B high-performance AI vision processor
- Quad-core ARM Cortex-A53 CPU
- Dedicated 3 TOPS NPU
- 12 MP at 30 fps ISP support
- 5 MP image sensor with 2592 × 1944 active resolution
- Up to 30 fps at full resolution in MJPEG format
- 4G+IR lens with F/2.2 aperture
- 6.8 mm effective focal length
- 52° diagonal, 45° horizontal and 29° vertical field of view
- Automatic gain, exposure and white balance
- 85 dB dynamic range
- USB 2.0 connectivity and bus power
- H.264 and H.265 video processing support
From Image to Checkout in Three Steps
Place
Place the tray or products inside the designated recognition area.
Recognize
The smart camera captures the scene while the AI model detects and identifies multiple items simultaneously.
Confirm and Pay
Recognized products and the calculated total appear on the checkout screen. Review the result, confirm the order and continue to payment.
AI Multi-Item Recognition Introduction
Watch the VideoBuilt to Work with Your
Retail System
The solution can integrate with existing POS and ERP environments, allowing recognition results to connect with product catalogs, pricing data and checkout processes.
Its modular design supports deployment as a dedicated recognition surface or as part of a self-service kiosk.
An optional second display can also be added to provide customer-facing order information or support different checkout configurations.
Ideal for Visual-First Retail
Self-Service Food Counters
Let customers place several dishes within the recognition area and process them through one streamlined workflow.
Bakeries and Donut Shops
Recognize products that may not carry individual barcodes and simplify the checkout of mixed trays.
Quick-Service Restaurants
Reduce repetitive manual entry and help staff keep checkout lines moving during busy periods.
