Conference paper Open Access
Probabilistic modelling combined with a CNN for boundary detection of carbon fiber fabrics
Sebastian Zambal; Christoph Heindl; Christian Eitzinger
Abstract:
For many industrial machine vision applications it is difficult to acquire good training data to deploy deep learning techniques. In this paper we propose a method based on probabilistic modelling and rendering to generate artificial images of carbon fiber fabrics. We deploy a convolutional neural network (CNN) to learn detection of fabric contours from artificially generated images. Our network largely follows the recently proposed U-Net architecture. We provide results for a set of real images taken under controlled lighting conditions. The method can easily be adapted to similar problems in quality control for composite parts.
| Name | Size | |
|---|---|---|
|
Zambal_2019_ProbModelCNN.pdf
md5:8941a55c64cb9e59e677863119063102 |
3.5 MB | Download |
| All versions | This version | |
|---|---|---|
| Views | 104 | 104 |
| Downloads | 136 | 136 |
| Data volume | 482.5 MB | 482.5 MB |
| Unique views | 103 | 103 |
| Unique downloads | 127 | 127 |
- Publication date:
- August 21, 2019
- DOI:
-
- Keyword(s):
- Deep learning, probabilistic modelling, U-Net
- Grants:
-
European Commission:
- ZAero - Zero-defect manufacturing of composite parts in the aerospace industry (721362)
- Meeting:
- IEEE International Conference on Industrial Informatics (INDIN'19), Helsinki-Espoo, Finland, 22-25 July 2019
- License (for files):
- Other (Non-Commercial)