Smart Grocery

Confidential

Confidential

Backend:
- Programming Language: Python, Linux
- Framework: RESTApi (main), Django (Gordon service)

Frontend:
- Programming Language: Javascript, Typescript
- Framework: Reactjs, Nextjs

Confidential

Confidential

Client is an Australian technology company specializing in AI-powered computer vision solutions aimed at revolutionizing retail operations. They focus on automating product recognition processes, particularly for items without barcodes, such as fresh produce and bulk goods.

Their technology enables faster, more accurate, and contactless checkout experiences, enhancing efficiency and reducing fraud in retail environments.

In retail, recognizing and classifying fresh produce and bulk items without barcodes is a common problem. These products often require manual entry at checkout, leading to slower transaction times, input errors, and increased risk of fraud. Such inefficiencies negatively impact both the customer experience and store productivity. Additionally, retailers often lack the necessary tools to effectively manage, configure, and monitor their product recognition systems. They need better ways to track performance and make updates.

SotaTek helped the client to deliver a custom AI solution that automates product recognition at checkout, focusing on items without barcodes such as fresh produce and bulk goods. Our team supported the development of advanced computer vision models and helped integrate them into the client’s scalable retail platform.

The AI system uses standard cameras to instantly identify products, removing the need for manual input and significantly reducing checkout time and errors. Additionally, the client’s Console and API platform enables retail partners to integrate, monitor, and customize the system to fit their operational needs.

Key solution highlights include:

• Automated Product Recognition: Utilizes computer vision to identify non-barcoded items instantly, eliminating manual entry and reducing checkout times by up to 90% for shoppers and 35% for cashier
• Fraud and Loss Prevention: Accurately distinguishes between similar-looking items ( preventing mislabeling and reducing shrinkage by up to 60%.
• Bag Detection: Detects the use of bags during checkout to apply correct tare weights, ensuring accurate billing.
• Hardware Agnostic Integration: Designed to retrofit existing retail hardware, allowing for easy and cost-effective implementation without overhauling current systems.

Key product features include:

• Recognition API: Accurately identifies non-barcoded items such as fruits and vegetables in real time, enabling faster and more efficient checkout.
• Management Console: A user-friendly platform for retailers to monitor system performance, configure settings, and analyze data insights powered by AI.
• Reporting & Metering System: Provides detailed tracking of system usage and performance metrics across different retail environments, supporting better operational decisions.

The client successfully launched an AI-powered retail solution that significantly improved checkout efficiency and operational accuracy. Retailers using the system have reported faster transactions, fewer manual entry errors, and a more streamlined in-store experience.