Two-Stage CNN-Based Wood Log Recognition

The proof of origin logs is becoming increasingly important. In the context Industry 4.0 and to combat illegal logging there an increasing motivation track each individual log. This work presents a two-stage convolutional neural network (CNN) based approach for wood log tracing on digital end images. First, cross section segmented from background by applying CNN-based segmentation method using Mask R-CNN framework. second step, recognition applied CNNs that are trained images triplet loss function. Our proposed achieves Equal Error Rates between 0.6 3.4% six employed image data sets clearly outperforms previous approaches recognition.