Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Smart Cameras: Using render image based object detection models in an Additive Manufacturing part sorting system
Sorting and counting parts is a tedious manual work during post processing for additive manufacturing. In this project, you will address this bottleneck using neural networks for object detection that are solely trained on synthetic training data.
Keywords: Deep learning, additive manufacturing, object detection, synthetic training data, image rendering, post processing automation, neural networks, powder bed fusion, part sorting
Smart cameras automatically collect information from images using computer vision algorithms. Actions and decisions can be triggered using this information.
Together with industrial partners, we are researching smart camera applications in the field of post-processing for Additive Manufacturing (3D-printing).
Printing parts using Powder Bed Fusion (PBF) requires manual sorting of the parts after fabrication and matching them to orders withs post-processing instructions. A system using smart cameras with an object classification model could automate this process by making use of the digital process chain.
Smart cameras automatically collect information from images using computer vision algorithms. Actions and decisions can be triggered using this information. Together with industrial partners, we are researching smart camera applications in the field of post-processing for Additive Manufacturing (3D-printing). Printing parts using Powder Bed Fusion (PBF) requires manual sorting of the parts after fabrication and matching them to orders withs post-processing instructions. A system using smart cameras with an object classification model could automate this process by making use of the digital process chain.
A pipeline to automatically train a neural network for image classification based on synthetic training data was realized and tested with industrial partners in prior projects. The synthetic training data thereby consists of image renderings created with the use of CAD files of the parts. You can find a video of the project here: https://www.youtube.com/watch?v=OrzP1kEgEZI
The task of this project is to adapt the pipeline and the existing system to increase its functionality. Key tasks are the following:
- Adapting the current system to realize object localization, for example by using object detection. This will require software and hardware adaptions.
- Using the information obtained from object localization to realize part counting.
- Evaluating the developed approach in an experiment.
A pipeline to automatically train a neural network for image classification based on synthetic training data was realized and tested with industrial partners in prior projects. The synthetic training data thereby consists of image renderings created with the use of CAD files of the parts. You can find a video of the project here: https://www.youtube.com/watch?v=OrzP1kEgEZI The task of this project is to adapt the pipeline and the existing system to increase its functionality. Key tasks are the following: - Adapting the current system to realize object localization, for example by using object detection. This will require software and hardware adaptions. - Using the information obtained from object localization to realize part counting. - Evaluating the developed approach in an experiment.
You are highly interested in working with neural networks and creating real world applications Desired are interest and skills in programming (Python preferred) and working with Blender Experience with the training of neural networks such as MobileNet, VGG-16, Yolo,.. will help You work independently and are eager to interact with an industrial partner You approach problems systematically and manage your projects methodically You are a team player, curious and come up with innovative ideas You are willing to integrate into a professional work environment
We focus on human-centred product development and regard the link between research and education as the key to excellence in training. We see ourselves as a partner for industry and promote the continuous transfer of knowledge through cooperation, as well as the training and further education of students and graduates to strengthen the competitiveness of mechanical engineering industry.
We focus on human-centred product development and regard the link between research and education as the key to excellence in training. We see ourselves as a partner for industry and promote the continuous transfer of knowledge through cooperation, as well as the training and further education of students and graduates to strengthen the competitiveness of mechanical engineering industry.
Bachelor/Semester Thesis Start: From February 2023
Jonas Conrad (pd|z)
LEE O 219
conradj@ethz.ch
Tel.: +41 (0)44 632 36 02
pdz.ethz.ch
Jonas Conrad (pd|z) LEE O 219 conradj@ethz.ch Tel.: +41 (0)44 632 36 02 pdz.ethz.ch