Trying the clothes to get the best look when buying from online shopping websites or apps.
What it does
This is an android application which allows you to try the clothes by using the images of the clothes and AR to place in onto the body in a camera preview.
How we built it
We used the images of the product which we get on sharing a product from online shopping apps. Then we processed the image of the cloth(product) to crop out the cloth out of the image. Then we used an open pose estimation model to find out the keypoints of the human body on to which the cloth is to be tried. Using the camera preview and a tensorflow model for open pose estimation we got those keypoints and placed the cropped image of the cloth on to it.
This app is built using Android Studio and is an android native app.
Challenges we ran into
One of the major issue was to get the image of the product. In online shopping apps, when we share a product we get a url of the product. We used that URL to download the image of the product by using web scraping using selenium. Also we are currently running the backend for the web scraping part on a local server.
Accomplishments that we're proud of
We have finally completed the first release of the app and it functions well and we can try any cloth from popular e-commerce websites like Myntra, Flipkart, Amazon and also we added Tchibo too.
What we learned
We learnt about how open pose estimation model works and also we learnt web scraping using selenium to get the image.
What's next for TRY-IT
Our next release is aimed at creating a 3-D model for the person using a single image and then place the clothes or try the clothes over that person virtually. A 3-D avatar resembling the user would make it easy for the user to style his outfits.