Image Analysis for Reading Out Embedden Information inside 3D printed Object using Deep Learning

This paper presents the method to extract the embedded information inside 3D printed object using convolution neural network (CNN). The information was embedded by printing the codes with the near-fluorescent filament as a thin layer inside the real objects using a FDM 3D printer. This code cannot be seen from the outside. To read out the information inside, we can use a specific near infrared CCD camera read out non-destructively. Because of the vary factors in embedding conditions and the lighting conditions in photography in practical use, the accuracy of decoding by only using brightness is inefficient. Hence, we use deep learning to solve the problem for automatic classification and read out better. The results from using CNN was very satisfied as the accuracy 96.8%.