The Visual Recognition model is capable of identifying over 2,000 different foods

IBM Watson has introduced a new beta feature that can identify different foods. The Visual Recognition food model provides enhanced specificity and accuracy for items. It is capable of recognizing over 2,000 foods in images.

watson-choc-brownie.jpg

Observing that users of food-logging apps get frustrated by the manual process of tracking their meals, the team came up with the idea of training a system to automatically identify foods in order to simplify food logging. Using the food model in the Visual Recognition API, Watson focuses specifically on the food shown in photos. This is different to general visual tagging, which identifies other items such as plates, knives, tables, and people.

Searching for images online, the team matched the correct foods to the correct labels to clean up the data set, and now they have the largest known collection of more than 1.5 million labeled food images corresponding to more than 2,000 different foods. They further developed a taxonomy around the foods to classify them hierarchically and to improve the system’s accuracy, they came up with a novel idea to exploit this food hierarchy in combination with deep learning methods for fine-grained recognition.

This food model has been developed with the restaurant, health and fitness, lifestyle, and travel industries in mind, making it simpler to automatically manage content and gather insights. It would let restaurant diners easily compare their meals to ones from previous visits to an establishment, and enable restaurants to better understand how often their food is being shared across social media. The food model can also empower review and delivery businesses to automatically classify both user-submitted images and official restaurant photos, allowing food-centric services to efficiently manage their content. The team is now excited to see the applications and interpretations developers and data scientists will build on their technology.

IBM Watson

IBM Watson has introduced a new beta feature that can identify different foods. The Visual Recognition food model provides enhanced specificity and accuracy for items. It is capable of recognizing over 2,000 foods in images.

Observing that users of food-logging apps get frustrated by the manual process of tracking their meals, the team came up with the idea of training a system to automatically identify foods in order to simplify food logging. Using the food model in the Visual Recognition API, Watson focuses specifically on the food shown in photos. This is different to general visual tagging, which identifies other items such as plates, knives, tables, and people.