Automated Knot Detection and Pairing for Wood Analysis in the Timber Industry

TL;DR

The paper addresses the automation gap in knot detection and pairing for timber processing by proposing a two-stage pipeline: (1) knot detection on high-resolution board surfaces using transfer-learned YOLOv8 models, achieving a maximum mAP@0.5 of 0.887, and (2) knot pairing via a triplet-network-embedded feature space followed by clustering, achieving 0.85 pairing accuracy with learnable feature weights. A large, preprocessed dataset of hand-annotated knots across four board surfaces enables robust detection, while a multidimensional knot-feature set (including k1, k2, and longitudinal coordinates) supports accurate pairing. The results demonstrate strong potential for industrial deployment, enabling faster, more reliable knot analysis and offering insights into which features most influence knot pairing in wood science. This work advances AI-assisted wood analysis and holds practical impact for improving efficiency and quality in the timber industry.

Abstract

Knots in wood are critical to both aesthetics and structural integrity, making their detection and pairing essential in timber processing. However, traditional manual annotation was labor-intensive and inefficient, necessitating automation. This paper proposes a lightweight and fully automated pipeline for knot detection and pairing based on machine learning techniques. In the detection stage, high-resolution surface images of wooden boards were collected using industrial-grade cameras, and a large-scale dataset was manually annotated and preprocessed. After the transfer learning, the YOLOv8l achieves an mAP@0.5 of 0.887. In the pairing stage, detected knots were analyzed and paired based on multidimensional feature extraction. A triplet neural network was used to map the features into a latent space, enabling clustering algorithms to identify and pair corresponding knots. The triplet network with learnable weights achieved a pairing accuracy of 0.85. Further analysis revealed that he distances from the knot's start and end points to the bottom of the wooden board, and the longitudinal coordinates play crucial roles in achieving high pairing accuracy. Our experiments validate the effectiveness of the proposed solution, demonstrating the potential of AI in advancing wood science and industry.

Figures (11)

• Figure 1: Two examples of knots on the different surfaces of the same wooden board. They are paired to each other.
• Figure 2: An illustration of knot pairing on a wooden board. The image to the left is a real-world photograph showing the actual knot pair on a wooden board. The diagram to the right is a cross-section that shows how the two knots are interconnected, forming a pair that appears on adjacent surfaces of the board.
• Figure 3: The proposed pipeline for knots detection and pairing. The workflow consists of two main stages: Knots Detection (the top section) and Knots Pairing (the bottom section).
• Figure 4: Example of images from the dataset.
• Figure 5: Explaination of 'K1' and 'K2' values.