Yuxiong Zhao (Jason)

1.Pneumonia X-Ray Image Detector using Convolutional Neural Network
(Personal Project)

What?

• Trained a typical Convolutional Neural Network (CNN) for pneumonia detecting.• Developed a Diagnostic Interface web API using Google Cloud Platform (GCP) allowing user to upload X-ray image to test for pneumonia using the model.

How?

• Preprocessed the labelled datasets.• Performed data augmentation• Traned a Convolutional Neural Network using labelled datasets. The trained and test accuraccy are 86.77% and 83.01% respectively.• Compared the model with VGG16 using transfer learning.• Created a Diagnostic Interface flask web API using Google Cloud Platform

2. Virtual Pen
(Personal Project)

What?

• Created a virtual painter allows users to virtually "draw" in the air with different color pens recognition.

How?

• Real-time video capture and image processing using C++ and OpenCV library, such as color recognition and transformation.• Used image enhancement algorithms to identify of specific colors by doing experiment to test color ranges, followed by contour drawing to pinpoint object positions.• Interface optimization like real-time horizontal flipping, scaling of the video stream etc.

3. The Facial Skin Thickness Detector
(Personal Project)

What?

• The skin thickness detector designed for aesthetic dermal filling industry

How?

- Developed the skin thickness detector using Arduino, force sensor, and calibrated servo robotic arm- Features extraction from the load--displacement profile.- Determined the result through the multivariate nonlinear regression- The results present a low-cost skin thickness estimate system with medical accuracy
- The prototype and the results were sold for further development

4. Design of the Autopilot
(University Project)

What?

• Designed an autopilot and tested on X-Plane

How?

• Built the Dynamic Model using MATLAB Simulink from the inertia and aerodynamic data.
• Built the autopilot and tuned the PID using the Dynamic Model.
• Tested the autopilot using X-Plane and compared the results with Simulink simulation.

5. The Altitude Controller of the Drone
(University Peoject)

What?

• The design of altitude controller & structural design of the drone

How?

• Calculated Predicted Dynamic Model
• Designed using MATLAB Simulink control model, with noise filter, and PID tuning
• Made reference to generative design, validated using FEA Abaqus

5. Control of a DC Motor
(University Lab)

What?

• Investigated the control properties of a DC motor to allow a precise control

How?

-Predicted the mathematical model and stability analysis
-Built the control model, filters, and circuit. Tuned PID
-Used Simulink Design Optimization to estimate parameters based on experiments.

4. Design of Wind Blade
(University Project)

What?

• Designed wind blade to generate maximum power for certain conditions

How?

• Derived aerodynamic model for maximum power generated
• Used MATLAB to generate coordinates defining the shape of wind blade
• Manufactured the blade with appropriate material according to material & structural analysis, and tested it in wind tunnel

7. Arial Robots
(Hobby 1)

Video 1: Optical flow & laser height positioning test

What?

• Drones built with innovative positioning methods like optical flow and laser positioning, object avoidance (as demonstrated by the video)
• Drones built with innovative 18650 lithium batteries for longer endurance.

How?

• Developed based on ArduPilot firmware, based on Arduino.
• Configured the overall design of the drone layout and circuit.

8. Design of RC Cars
(Hobby 2)

What?

• RC cars are designed for remote outdoor patrols, with a long control range and strong off-road capabilities.

How?

• CAD amorphous objects with high precision
• 3D printed and CNC manufactured components

Video 2: My car & cute wildcat (^ _ ^)