-
ex 1
-
ex 2
-
website view
Inspiration
Macy's challenge sounded like an interesting task, even though we both have full time job offers, we decided to put our minds together to tackle it!
What it does
Macy's stated that they only had a 4% conversion rate from a user entering their website to an ACTUAL sale. We realized that a reason for this may be that visitors who access the site may not actually come looking for a specific object, and rather are "window shopping". We had the idea of giving the user suggestions of retail items on Macy's while they were doing OTHER tasks. For example, the user is browsing their favorite fashion blog on tumblr or pinterest. Ask Macy's will analyze the images on the blogs in the background and notify the user of similar clothing/items that are available for sale on Macy's. This way the user is presented with items that are based on their browsing interests.
How I built it
Extension's core is built using angular.js, jQuery for asynchronous http requests, and javascript for core logic. Html and CSS were used for aesthetics. The Machine learning side is built off of Clarifai's Machine learning API. It enables developers to categorize images and videos into certain categories based on its learned knowledge base. We also extracted the colors from the images on our page to narrow our searches even further. We communicated with this API using asynchronous http requests in javascript.
Challenges I ran into
Dealing with the intricacies of angular.js and jQuery to make asynchronous requests. Developing a service that uses real time machine learning and classification in the browser was an extremely difficult task to do in >12 hours! Also having access to Macy's API would have saved hours of web scraping efforts.
Accomplishments that I'm proud of
Developing a real time machine learning classification pipeline and effortlessly displaying results to the end user. The extension also does not create any unnecessary load for the end user.
Applying machine learning to a real world application that has massive potential.
What I learned
angular.js, jQuery is still difficult, and learned alot about macy's catelog.
What's next for Ask Macy's
The service has massive potential in the real world. We'd love to share our work with Macy's and discuss details on further implementaton , perhaps using the Macy's API.
Log in or sign up for Devpost to join the conversation.