BRUSHY
Team members
Instructors:
-
Sun Zhu
-
Oka Kurniawan
Writing Instructors:
-
Belinda Seet
Hi, I'm Brushy!
Did you know?
Oral diseases affect 3.7 billion people around the world. Oral-B’s round head iO power brush has capabilities to significantly improve oral health.
Not everyone knows about this.
That's where I come in!
I simulate brushing movements of power and manual brushes, applying consistent pressure and showing you the differences in plaque removal!
That way, you can see the benefits of power brushes in real-time and make better choices for your oral health.
See me in action!
User journey
User feedback
Impact
Main subsystems
Brushing Subsystem Concept Iterations
Swipe to see more!
Brushing Subsystem Prototype Iterations
Plaque Visualization Model Architecture
Overview
Our plaque visualization model employs image segmentation algorithms to accurately identify and visualize plaque simulant regions. This allows for precise demonstration and evaluation of cleaning efficiency.
Model Workflow
Feature Extraction with CNN
Images (support and query) are processed through Convolutional Neural Networks (CNNs) to extract low-resolution feature maps crucial for distinguishing plaque simulant from typodont surfaces.
Prototype Generation
The model utilizes a self-support mechanism to refine the mask by identifying high-confidence plaque pixels within the probability map of the initial query mask, aggregating these features into a self-support prototype.
Similarity Matching
Cosine similarity was used as a metric to measure the closeness between the query image features and generated prototypes, accurately determining plaque simulant regions.
Segmentation Output
Overall System Architecture
