Food Recognition Technology for Nutrition and Wellness Apps

Start building awe-inspiring apps today that recognize thousands of types of food down to the macronutrients level!

Get started today

Data Scientists at BroutonLab spent a significant of time feeding data into AI algorithms and perfecting an image classification model on food.

We trained neural networks to recognize thousands of different foods down to macronutrients level. This model can tell how healthy (or unhealthy) your meal from a picture.

Try it out now

Select picture

Characteristic

r1
r2
r3
r1

Characteristic

Why To Use Food Recognition Model?

If you are working on health, fitness, lifestyle, or any food-centric app, food recognition model is a must-have to increase app downloads and boost user reviews.

Review platforms like OpenTable, Seamless, Foursquare can automate the categorization of user-generated content and manage images submitted by restaurants. It reduces manual work and money required for media management!

Restaurants and consumer brands can increase their Marketing ROI by gaining deeper insights into what dishes are most photographed and shared by their customers on social media. So re-targeting content and tailoring it to the audience gets super easy!

Questions? Ask Us Now!

How Did We Develop This Food Recognition Model?

Our Data Science team used a convolutional neural network’s base to generate features from the image, and the classifier is to classify the image based on the detected features. We then added a new classifier to fit our goals and fine-tuned our model according to one of three strategies.

With the Sigmoid activation function we converted each score of the final node between 0 to 1 independent of what the other scores are.

We e xperimented with all types of Resnets, Mobilenets, and even Nasnets. For further development, we picked a lightweight Mobilenetv2 and NASNetMobile with accurate Xception and InceptionResNetV2 models.

Talk To Us Now