Projects

Applied engineering work spanning camera control, industrial automation, and deep learning.

Hardware-Software Integration, Industrial Automation & Deep Learning Projects

Browse a portfolio of applied engineering work combining camera control software, industrial automation, and deep learning. Every project below is a working system built to tackle a genuine measurement, classification, or imaging problem.

Control software synchronizing a color camera and a 3D profiling camera on a conveyor belt
Synchronized control of color and 3D profiling cameras
Ore sorting system fusing color and hyperspectral imaging for rock classification
Ore sorting with color and hyperspectral imaging
Satellite terrain classification across city, sand, vegetation, mountain, and water classes
Satellite image terrain classification
Anthropometric body measurement estimation using OpenPose keypoints
Anthropometric measurement estimation with OpenPose

Applied Research & Systems

Control Software for Color and 3D Profiling Cameras

Hugo Negreyros and David Shatwell built software that controls and synchronizes two Gigabit Ethernet cameras simultaneously: a Teledyne Dalsa Spyder 3 color camera and an Automation Technology C2-2040 3D profiling camera. The setup enables high-speed characterization of materials traveling along a conveyor belt.

Code

Ore Sorting using Color and Hyperspectral Imaging

David Shatwell, Victor Murray, and Augusto Barton developed an autonomous method that combines color, texture, and hyperspectral features to classify mineralized rock. The work reached an MCC of 0.9787 and was selected as one of 25 finalists at the PERUMIN 35 National Mining Contest.

Classification of Satellite Images based on their Class of Terrain

David Shatwell, Alejandro Weston, and Oscar Ramos created a terrain classification system that separates five classes — city, sand, vegetation, mountain, and water — by pairing deep convolutional architectures with classical feature classifiers on Google Earth imagery.

Code

Anthropometric Measurements Estimation using OpenPose Algorithm

David Shatwell designed an automated medical triage system that relies on OpenPose body keypoint coordinates to estimate frontal projections and body circumferences for clinical occupational screening.

Code

Frequently Asked Questions

What kinds of projects are featured on this page?
The projects span hardware-software integration, industrial automation, and deep learning implementations — including camera control software, ore sorting with color and hyperspectral imaging, satellite terrain classification, and anthropometric measurement estimation.
Which projects include publicly available source code?
Several projects offer a public repository via a "Code" button: the color and 3D profiling camera control software, the satellite image terrain classification project, and the OpenPose-based anthropometric measurements project.
What imaging hardware is used in the camera control project?
The system controls and synchronizes two Gigabit Ethernet cameras at once: a Teledyne Dalsa Spyder 3 color camera and an Automation Technology C2-2040 3D profiling camera, enabling high-speed characterization of conveyor belt materials.
How does the ore sorting project classify rock samples?
It autonomously combines color, texture, and hyperspectral features to classify mineralized rock, and was chosen as one of 25 finalists at the PERUMIN 35 National Mining Contest.
Where can I find related publications?
Related research appears on the Papers page, and more code is available on the GitHub page.

Papers GitHub

© 2026 David Shatwell

Each project on this page reflects a hands-on approach to solving real engineering problems, from synchronizing industrial cameras to applying deep learning for classification and measurement. Use the Code buttons to explore the public repositories, or visit the Papers and GitHub pages for related research and additional work.