AI for recognizing dishes on the hotel's Swedish line

The company ItFox collaborates with the large restaurant chain London Restaurant Group. Therefore, it was necessary to develop an automated solution capable of speeding up the process of dish issuance.

Implementation:
To solve the task, ItFox decided to develop an artificial intelligence-based system capable of automatically recognizing dishes and displaying their names on the "Fotobox" screen. This was expected to significantly simplify and speed up the issuance process, as well as reduce the likelihood of errors related to the human factor.

Key implementation steps:
1. Data collection and analysis of existing solutions. Since no suitable solutions were found on the market, it was decided to develop a proprietary system.
2. Development of a service to prepare data (photos of dishes and their names) for transmission to the neural network.
3. Development of a service for dish recognition using a highly accurate neural network. This required addressing issues related to performance, recognition quality, and image preparation.
4. Creation of functionality to automate the model training process.
5. Development of a mechanism for delivering the trained model to the "Fotoboxes."
6. Optimization of the training process to achieve maximum speed and recognition accuracy.
7. Implementation of the system, including staff training, addressing organizational issues, and optimizing the placement of equipment.

As a result of implementing the AI system, the speed of dish recognition significantly increasedto 1.5-2 secondsand the recognition quality improved to 90-95%. This allowed for faster customer service and simplified the staff's workflow. The system updates automatically, maintaining the relevance of recognition models.