Robot YOMY
Our Computer Vision models embedded in a smart dispensing robot
98%Model precision
350Inference time in milliseconds
3Trained Models
The Challenge
- YOMY is a French start-up developing a robot dispenser for cat kibble and wet food, equipped with an embedded camera.
- The company wanted to integrate image analysis technology enabling it toidentify different cats within the same household, day and night, in order to adapt meal frequency and quantity to their specific needs, while offering owners reliable and automated nutritional monitoring.
Our Solution
- Galadrim developed and trained several deep learning models, including an image classifier and a detection model based on convolutional neural networks (CNNs). An identification model combining a CNN backbone and a Siamese network was also designed to accurately differentiate the most similar-looking animals. The models were embedded on the robot, with optimised hardware sizing and in-depth inference testing, ensuring high performance in real-world conditions.
