Einzelansicht

Title: Real-Time Embedded Vision and Machine Learning
Short Code: EVA_EVaML
ECTS Credits: 2
UAS: ZHAW
Organizer Details: MRU ZHAW Institute of Embedded Systems
Evaluation:

pass/fail based on results of programming assignment

Decision Date: 10. März 2023 
Start Date: 10. April 2023 
End Date: 26. Mai 2023 
Date Details:


Type:

Half day courses on 7 days in the semester plus individual work on assignment

Language(s):

English

Description (max. 300 characters):

The module targets the field of embedded real-time vision systems with AI support. Such systems are typically used for automotive, drones or industrial applications. The students get an introduction to state-of-the-art embedded camera and processing technologies. Each student gets an embedded vision kit based on a raspberry pi, which will be used for hands-on experiments. The full development chain from embedded Linux, camera driver integration, AI tools, neural network training, integration and a real-time application will be practiced.

Contents and Learning Objectives:

The course is setup as a 7 half-day workshop. Every student receives a raspberry-pi and a camera. The students will develop individual demos (e.g. industrial object recognitions, tour guides, collision alerts). All demos will be discussed and reflected during the workshops. The specific goals:

  • Training neural networks for small (mobile) devices in order to gain knowledge of the real time behavior of neural networks used for vision-based systems.
  • After investigating popular datasets such as MNIST, the students will elaborate specific use cases, e.g. detection of specific objects with a camera.
  • The full development and processing chain will be defined, such as data acquisition, labeling, training, testing and porting the neural network to the target.
  • The real time capabilities of the target system will be analyzed and tested.
  • The following tools will be used: Keras, Tensorflow, Python, and open source code generation applications for converting neural networks into C.
Admission: ET, IT, ST
Literature:
Conditions:

50% theory, 50% lab work

Contact:

Prof. Dr. Matthias Rosenthal, +41 58 934 78 39

 
Contact Person E-Mail: rosn@zhaw.ch
Status: registration open
 
Specialization: Computer Science (CS)

Data Science (DS)

Electrical Engineering (ElE)

Mechatronics & Automation (MA)

 

[Responsible for this text: Kuzmanovic-Tesic Jelena]