© 2025 Advanced Research and Training Academy (ARTA). All rights reserved.
Certificate
Machine Vision on the Rock 5B SBC
Course Outline
The Machine Vision on the Rock 5B SBC course provides advanced training in accelerated embedded AI systems using the Rock 5B development board powered by the Rockchip RK3588 processor.
Learners will work with an 8 TOPS Neural Processing Unit (NPU) to deploy high-performance deep learning inference models on embedded hardware. The course covers model conversion from PyTorch to RKNN format, optimisation for hardware acceleration, and deployment on real-time video streams.
Students will train state-of-the-art object detection models such as YOLOv8 and YOLOv11, develop custom OCR pipelines using EasyOCR, and deploy trained models to embedded systems for real-world applications.
By the end of the course, learners will design, train, optimise, and deploy complete accelerated machine vision pipelines on advanced Single Board Computer (SBC) hardware.
Key Information
- MQF/EQF Level: Not pegged/recognised by MFHEA
- ECTS: N/A
- Qualification/Certification Type: Certificate of Completion
- Course Duration: 20 weeks × 3 hrs/week
- Total Hours: 60 hours
- Delivery Mode: In-person (hands-on lab-based)
- Language of Instruction: English & Maltese
- Target Audience: 16+ years
Students must complete at least 85% of total learning hours (tracked via attendance logs, lab participation and assessment submissions). A Certificate of Completion will be issued accordingly.
Course Description
This unit builds directly on the knowledge and skills developed in the prerequisite module Programming Python for the Raspberry Pi. Students will work with the Rock 5B development board, powered by the Rockchip RK3588 octa core processor, one of the most advanced and high-performance CPUs available in modern Single Board Computers. The board features four 2.4 GHz cores, four 1.8 GHz cores, 16 GB of onboard RAM, and an integrated 8 TOPS Neural Processing Unit (NPU) designed to accelerate deep learning inference.
Students will learn how to convert PyTorch models into the RKNN format to leverage the hardware NPU, achieving inference speeds up to ten times faster than CPU only execution. The unit also covers custom training of state-of-the-art machine vision models, including YOLOv8 and YOLOv11, using domain specific datasets. In addition, students will explore custom training workflows for EasyOCR to perform optical character recognition on bespoke datasets.
All trained models will be deployed on the Rock 5B and tested using real time video streams from USB/CSI cameras, IP cameras, or still images. By the end of the unit, students will integrate these concepts into a practical case scenario project, demonstrating full end to end implementation of accelerated machine vision pipelines on embedded hardware.
Entry Requirements
- Successful completion of Programming Python on Raspberry Pi
- Strong Python programming skills
- Basic understanding of machine learning concepts
Recommended:
- Familiarity with:
- OpenCV
- Linux terminal usage
- Basic neural network concepts
- GPU/NPU acceleration basics
This course is not suitable for beginners.
Target Audience
This course is designed for:
- Embedded AI developers
- Robotics engineers
- Computer vision developers
- Automation engineers
- Mechatronics students
- AI/ML practitioners transitioning to embedded deployment
- Advanced ICT students specialising in AI systems
Course Delivery
The programme is structured into eight modules, delivered through:
- Instructor-led demonstrations
- Guided AI model training workshops
- Embedded deployment labs
- NPU acceleration configuration sessions
- Dataset preparation exercises
- Real-time video testing
- Capstone project development
Course Structure
- Module 1 – Rock 5B Architecture & NPU Fundamentals
- Module 2 – PyTorch Model Preparation
- Module 3 – RKNN Toolkit & Model Conversion
- Module 4 – YOLOv8 & YOLOv11
- Module 5 – Optical Character Recognition
- Module 6 – Real-Time Video Integration
- Module 7 – Embedded AI Optimisation
- Module 8 – Final Capstone Project / Final Assessment
Assessment & Certification
Assessment Breakdown:
- Practical Lab Assessments – 40%
- Model Training & Deployment Assignments – 20%
- Mid-Term Technical Evaluation – 10%
- Final Capstone Embedded AI Project – 30%
Upon successful completion, students receive a Certificate in Machine Vision on the Rock 5B SBC.
Learning Outcomes
By the end of this course, learners will be able to:
- Train custom deep learning models using PyTorch.
- Convert Pytorch models to ONNX and later to RKNN.
- Deploy accelerated inference pipelines on embedded hardware.
- Train and deploy YOLO object detection systems.
- Implement custom OCR systems using EasyOCR.
- Optimise real-time video inference performance.
- Design complete end-to-end embedded AI systems.
- Troubleshoot hardware–software performance bottlenecks.
Course Intake Dates & Pricing
April 2026
Price: EUR 1300
Optional Add-ons (please contact us for pricing):
- Rock 5B (16GB) + Power Supply
- Full AI Kit (Board + Camera + Storage + Accessories)
Advancing Education. Empowering Innovation.
Sign Up for Course Alerts
Get notified when new courses launch so you never miss a learning opportunity.