The AutoRate Restaurant experience is inspired by the latest generation of social communication tools, like Snapchat, which allow users to share events in their lives without the need for words.

What it does

AutoRate Restaurant allows users to submit a rating for a restaurant by uploading a selfie (or a group photograph) of them enjoying their dining experience. The application automatically analyses the photo to detect emotions in the image, which are then converted into an overall rating of the user's experience using Microsoft's Cognitive Services APIs.

The user isn't required to spend any time carefully wording their review and can take a selfie & upload it to the website from their smartphone in seconds.

Detected emotions & confidence levels are then analysed by our application, and computed into a final score using an algorithm tailored for food & hospitality contexts.

How we built it

AutoRate Restaurant is built using the Yii PHP MVC framework & makes use of Microsoft's Cognitive Services APIs for emotion detection. The application uses a simple LAMP stack with an Nginx reverse proxy for the production deployment to serve the uploaded images & static assets.

Challenges we ran into

The most significant challenge faced when developing AutoRate Restaurant was developing an algorithm which could appropriately analyse the emotion scores outputted by the Microsoft service and compute a final rating for the restaurant.

What we learned

Developing the application allowed us to experiment with new APIs, the group had no prior experience with Microsoft's Cognitive Services APIs and developing the AutoRate Restaurant application gave us an opportunity to experiment with the API's potentials.

The project also enabled the team members to train each other in their areas of expertise. This in particular allowed the team to share knowledge about different frameworks. Only one team member had used the Yii framework previously, likewise it was a good chance to share collaboration skills, including Git repo management techniques.

What's next for AutoRate a Restaurant?

We envision that the AutoRate Restaurant application server as a good proof of concept for further development. It would be possible to apply similar techniques used by the application to a range of different tasks, such as automatically analysing the content of images posted at events on social media in order to provide a review of the event without the usual cognitive bias associated when analysing text-based reviews.

Built With

Share this project:


posted an update

We were 10 minutes short of releasing this feature in time for the hackathon submission but we couldn't let our app be unfinished for long, so immediately after recovering from the jet-lag we completed & released geolocation functionality to find and rate nearby restaurants.

Log in or sign up for Devpost to join the conversation.