FaaiLok: An AI-Powered App for Children to Craft
Personalized Chopsticks

Project Background

Customization

  • The customization market has grown significantly, driven by consumer demand for personalized products.
  • The children’s customization market is still underdeveloped, presenting an opportunity for growth.
  • The lack of standardized methods in customization poses challenges for quality and efficiency.

Traditional Culture

  • Cultural heritage represents the collective knowledge of a society
  • Traditional cultures are at risk of disappearing, often due to ineffective propaganda
  • Innovative approaches like customization can make learning about culture more engaging and relevant, allowing individuals to express their creativity while connecting with cultural elements.

Advanced Technology

  • Pre-trained text-to-image models are increasingly researched for their customization potential.
  • Technological advancements, particularly in artificial intelligence, have revolutionized content creation and customization.
  • Generative models enable users to produce unique artworks and designs efficiently.

Children’s Education and Happiness

  •  there is a strong societal focus on academic achievement, often leading to an exam-oriented education model
  • This approach prioritizes academic success over creativity and happiness, hindering children’s personal growth
  • The pressure to excel can stifle imagination and joy, necessitating methods to help children regain a sense of happiness and creativity while aligning with the existing educational system.

Project Components

An interactive application for children to have fun


An application shall be developed for children to design effortlessly. It will have a user-friendly design interface, supporting digital and traditional drawing methods: children can both draw directly in the drawing panel of the application on tablets and draw on paper to scan and upload.

AI generative models to enable standardized customization


AI generative models should be implemented to convert children’s drawings into intricate artworks with multiple artisitc style selections. This feature will utilize deep learning algorithms trained on various traditional designs and child’s drawings to ensure the output remain artistic styles and the individuality of each child’s artworks.

3D modelling function for fine production


There should be a detailed 3D modelling function to allow users to visualize their designs in an authentic manner. Once a child’s drawing is transformed into a carving, the application will generate a 3D model, offering an interactive experience that allows users to manipulate their designs through rotation and zooming.

Actual Project Timeline


Meet with Supervisor +
draft detailed project plan

AI generative framework development

Mid-Project Presentation

3D modelling function development

Application development

Testing and Refinement

Final Optimization

Final Presentation(April-May 2024)

September 2024

November – December 2024

January 2025

Mid-March 2025

Early-April 2025

April 2025

April 2025

April-May 2024

Methodologies

1. Application Development

2. AI Generation Development

Minor Details:

A. 500 artworks collected for Chinese-ink style LoRA model training, each labelled by CLIP-Interrogator. Further, two other models are incorporated, which are listed below.

B. Chinese-ink LoRA model generation results

C. White-and-black porcelain style LoRA model generation results

D. Disney style LoRA model generation results

3. 3D Modelling Function

Generation Results:

B. Toy Generation (utlize Stable Fast 3D model)

Generation Results:

Deliverables

1. Detailed Project Plan

2. Interim Repor

3. Final Report