Visual Anomaly (Defect) Detection in LabVIEW
This is a demonstration of a unsupervised deep-learning based visual anomaly(defect) detection implemented in LabVIEW using the PatchCore method.
By utilizing DeepLTK (Deep Learning Toolkit for LabVIEW) and CuLab (GPU Toolkit for LabVIEW) for post-processing, we can train new models in under a minute and achieve inference times as low as 10ms.
More details on implementation in blog post:
https://www.ngene.co/post/deepltk-tut...
Project on GitHub:
https://github.com/ngenehub/deepltk_e...
PatchCore Paper:
https://arxiv.org/abs/2106.08265
DeepLTK (Deep Learning Toolkit for LabVIEW):
https://www.ngene.co/deep-learning-to...
CuLab (GPU Toolkit for LabVIEW):
https://www.ngene.co/gpu-toolkit-for-...