HackOn with AMAZON
Ecommerce has been able to rake in the kind of popularity that it’s experiencing today all because of the fact that it without fail delivers an engaging shopping experience in the comfort of our homes our workplaces, you name it! In the times of ongoing worldwide pandemic, when isolation is the key to not spread the virus, we wanted to enable customers to get as close as possible to the experience of shopping in a store while they are very much in their own comfort zones. As shoppers and most of us living by ourselves, we’re all guilty of indulging big time when taking impulsive decisions while shopping on our own. That’s where the advice of family and friends is much needed. It is definitely one most loved experience we’re missing out on while shopping online.
The concept of Collaborative Shopping with AR tends to bridge that very gap.
Broadly speaking, the end goal of our project here was to implement a virtual try-on network — essentially taking in-shop clothing and a person image as input to give output as an image of a person wearing those clothes and providing an in-built social chat application to enable users shop along with their family & friends with the comfort of being at home (social distancing in these dare times). The model implemented has been done with a focus on tops, with complete apparel transfer being potential future work.
Model Used: Down to the last Detail Virtual-Try-On
Working:
STEP 1: Model Image is processed using OpenPose to mark keypoints
STEP 2: Using the keypoints, the image is segmented into various features: face, arms, body etc.
STEP 3: The cloth mask of the clothing item is generated.
STEP 4: The mask and the segmented image are warped together.
- Current Key Features:
- User can Login/Sign Up
- Email verification for first time users
- User can also order without creating an account and logging in (Guest checkout feature)
- Payment feature through Paypal API
- User can add multiple number of items of a same product
- Order summary is visible after payment confirmation
- Logged in user can have multiple shipping addresses
- Logged in user can see how a particular dress looks on them
- User can make groups and add family & friends
- Groups have a shared cart and each member can add items to that cart
- Future Scope:
- Groups will have a chat feature where the members can talk and take each others feedback on every product
- Providing a real 3-D environment for an immersive and fun shopping experience
- Allowing customers to try on and assess different products virtually so that they can decide how a dress, necklace, or lipstick shade would look on them, or if a piece of furniture would look good in their bedroom.
Clone this Repository on your local machine by running the following command on your terminal -
git clone "https://github.com/hackko-20/Virtual-Try-On.git"
OR download the zip folder.
Install Anaconda3 on your PC, and add it to your path. Follow the steps below in your terminal:
cd "Path where you have cloned this repository"
Create a virtual environment in conda
conda create -n tryon
conda activate tryon
pip install -r requirement.txt
To run server, run the following command:
python manage.py runserver
For the model, clone Repository. In Virtual-Try-On/store/views.py under the function "virtualTryOn" :
(line 130) Change the path to the current path of Down-to-the-Last-Detail-Virtual-Try-on-with-Detail-Carving\demo\demo.txt on your computer.
(line 141) Change the path given in subprocess to the current path of Down-to-the-Last-Detail-Virtual-Try-on-with-Detail-Carving/demo.sh on your computer.
In Down-to-the-Last-Detail-Virtual-Try-on-with-Detail-Carving/demo.sh:
(line 4) Change the path to the current path of Down-to-the-Last-Detail-Virtual-Try-on-with-Detail-Carving on your computer.
(line 5) Change --forward_save_path value to the current path of Virtual-Try-On/static/images/Viton on your computer.
- Website Frontend: HTML, CSS, BootStrap, JavaScript
- Website Backend: Python Django Framework, Paypal API for payment, Web Socket (Django Channels)
- Model Backend: Python, MATLAB, Ananconda3, Libraries (Pytorch, Scikit-Image, OpenCV).
With more and more customers preferring to shop from retailers offering virtual reality eCommerce stores, the popularity of VR-enabled eCommerce platforms is increasing by leaps and bounds across the globe. And with the ongoing worldwide pandemic, it has become the need of the hour.
We have used these emerging technologies in our project that can take customer experience to a whole new level by:
- Facilitating tech-enabled shopping
- Helping virtual retailers transcend geographic barriers
- Allowing customers to try on and assess clothes virtually so that they can decide how a dress looks on them
- Keeping customers engaged
- Educating customers about your products
- Saving customers’ time.
- Allowing customers to have realtime feedback of how a dress looks on them from family and friends
It would also help the retailers by:
- Enhancing customer satisfaction
- Reducing the number of returns
- Increasing online retail sales and
- Boosting conversion rates.