-
(1/3)Top-left: Your recommended shoe. Other 3: Amazon-sourced matches based on features.
-
(2/3)Top-left: Your recommended shoe. Other 3: Amazon-sourced matches based on features.
-
(3/3)FALLBACKTop-left: Your recommended shoe. Other 3: Amazon-sourced matches based on features.
About the Project
The project was inspired by the challenges faced by disabled individuals when shopping online. Traditional e-commerce platforms are often overwhelming and inaccessible, making it difficult for them to find products tailored to their needs. This inspired us to create an application that simplifies and personalizes the online shopping experience.
What We Learned:
The importance of accessibility in technology design. How to leverage Natural Language Processing (NLP) to interpret user input effectively. Building a user-friendly interface with intuitive workflows and 3D rendering to make interactions seamless.
How We Built It:
NLP Engine: Interprets user input to extract preferences like shoe type, color, and style, making the process intuitive. Database Integration: Connected to a robust dataset of 35,000 shoes for detailed filtering and matching. Backend System: A fallback mechanism ensures relevant recommendations, even with vague inputs, while a Random Forest Regressor (RFR) provides scalable and ranked suggestions. Amazon API: Links users to similar products for easy purchase. 3D Visualization: Lightweight rendering tools allow users to explore customizable shoe designs interactively.
Challenges We Faced:
Fine-tuning the NLP engine to handle vague or incomplete inputs while maintaining high accuracy. Selecting a scalable model: transitioning from simpler ranking systems to implementing a Random Forest Regressor for better performance and future scalability. Efficiently managing a large dataset without compromising response times. Creating a lightweight yet realistic 3D rendering system to enhance accessibility for disabled users.
Future Plans:
We aim to improve the Random Forest Regressor by training it with real-time user feedback, making recommendations smarter and more accurate. In the future, we’ll move beyond ranking to a full AI-powered system for truly personalized suggestions, ensuring a better experience for every user.
Log in or sign up for Devpost to join the conversation.