Midway Design Review Team 1: MirrAR Team Members: Taehwan Chon, Jyerrmee Nerona, John Nguyen, Kevin Pierre Advisor: Russell Tessier
MirrAR Team Taehwan Chon John Nguyen Jyerrmee Nerona Kevin Pierre
Problem Statement Limited space for inventory Major transition to ecommerce and digital marketing Desire for precise, real-time AR display
Facial Recognition Requirements and Specifications App should provide 90 degrees of motion Image display delay <1 second 3D Models should display correctly on users face Competent scaling ability Bluetooth communication between hardware and software for additional landmark/tracking Operational range of <5 feet
Glasses Frames Requirements and Specifications Provides accurate measurement of face Will add additional landmark to detect facial angle Safe, comfortable and easy to operate Will be lightweight and balanced Cheap to manufacture Durable
Block Diagram
MDR Deliverables Android Application (David) Facial recognition and tracking (David and John) Demonstration of application communication Android application displays image of glasses on MirrAR (David) Hardware Design (Jyerrmee) Circuit Design (Kevin)
Android Application and UI: Taehwan Android Studio Integrated Development Environment (IDE) Android 6.0 Marshmallow Software Development Kit (SDK) Tools: Resources provided by Android Studio to enable specific development and features ASUS Zenpad 3S 10
Communication Between Hardware and Software HC-06 Wireless Bluetooth Module BluetoothAdapter Allows you to perform fundamental Bluetooth tasks Camera android.hardware.camera2
Facial Recognition Software: John Utilizes OpenCV library package CV library comes with classifiers that are trained to detect face Able to detect features such as eyes, nose, face, mouth Displays green rectangle around face once found
How Does Facial Feature Detection Work? Determine if there is a face in image provided, by applying classifiers Does this by checking values of pixels Once face detected, individual features are analyzed If no face, does not proceed further
Software Used Software ran on Netbeans to test program with a use of camera Uses package “gui” to simulate an interface Panel for camera image display Start button Pause/Stop button
Pseudo Code Class detectFace{ -Activate device camera -Detect face in live feed and store in memory -Implement classifiers into code -Once facial features detect, draw rectangle around face }
Hardware Design: Kevin Pierre Attiny85 8 kb Flash memory program storage 20 MIPS at 20 MHz Operating Voltage: 2.0 - 5.5 V HC-06 Bluetooth Module Follows IEEE 802.15.1 standardized protocol Range: Less than 100m Operating Voltage: 2.0V - 6V
Power Requirements Current Draw: Attiny85: 10mA HC-06: 15mA If using an CR2032 battery, with 240mAh
Power Requirements
3D Design: Jyerrmee Reference Frames V1 CAD Software: Fusion 360 3D printed (Plastic): Easy assembly Fits majority of adult size heads Comfortable adjustable ear hooks Safe design: Eye hazards Operation Adjustable pieces are friction fit to its slots: Hold measurement Alternative nose mount design
Initial Dimensions Head width compatibility = 144mm to 164mm Avg human head breadth = 148mm Range of adult human head breadth = 140mm to 165mm Most frames have temples that range in length from 120 to 150 mm *For both male and female*
Weight Battery + Battery Holder: 15 grams PCB: 10.5 grams 3D Printed frames: 26.4 grams Total Weight: ~51.9 grams Typical sunglasses weigh up to 50 grams
Reference Frames V2 + PCB + Pots
Tablet Selection ASUS Zenpad 3S 10 Android 6.0 Marshmallow 9.7” 2K IPS Display (2048 X 1536) Hexa-Core Processor Dual Core ARM Cortex 2.1 GHz Quad Core ARM Cortex 1.7 GHZ, 64 bit MediaTek Processor Webcam Resolution Rear Cam: 8MP Front Cam: 5MP
Proposed CDR Deliverables Bluetooth communication (David & Kevin) Complete Design (Jyerrmee & Kevin) Custom PCB (Kevin) All potentiometers (Jyerrmee) Safe & Comfortable housing (Jyerrmee) 3D Models (David) Additional landmarking/tracking (John)
Gantt Chart
Demo Demo
Thank You Questions