The client needed an automated inventory system, which we build with Vision AI. The client has numerous vending machines and market shelves that require constant monitoring and refilling. This necessitates a significant amount of manual effort. We automated the inventory process by training a Vision AI model to detect empty spots and the products on the shelves, resulting in immense cost savings.
Challenge
As a market leader in hospitality, our customer faces the challenge of monitoring tens of thousands of machines and shelves. The sheer scale – monitoring vending machines and market shelves in the tens of thousands – demands significant manpower. Human oversight is inherently slower and more prone to errors compared to an inventory automated with AI. Updating inventory lists on a regular schedule is time-consuming. Ideally, a real-time inventory system would enable proactive restocking, preventing negative customer experiences and ensuring product availability.
Solution
We implemented an inventory monitoring automated with AI with a cost efficient Raspberry PI based camera system and custom-built software. Our model was trained to identify empty spaces and common products on market shelves. These Vision AI models are hosted in the Google Cloud with GPU support for high performance.
Result
The rollout of our AI automated inventory system is still in progress. Sites using the system are saving significant amounts of money. Furthermore, they are speeding up the restocking process, which improves revenue and customer satisfaction.