XyloTron and XyloPhone

Photo Credit

Forest Service photo

A researcher demonstrates the use of a XyloTron to image and identify wood samples.

Screening of wood products at ports, border crossings, or other points of control is largely done using a hand lens to identify wood anatomical features. Screening by human experts is limiting in the fight against illegal logging because too few people are adequately trained to perform this task. To make this process faster and less dependent on individual staff expertise, Forest Products Laboratory researchers and collaborators have been conducting research on computer vision-based methods for wood identification and developing technologies to deploy these methods in the field. The objectives of such field screening are to rapidly move legal products into trade and to establish probable cause to seize and forensically analyze any products in violation of laws or regulations.

The XyloTron is a portable wood identification system in which a XyloScope, a custom-designed imaging device, is used to take magnified pictures of wood samples. The XyloTron captures images showing wood anatomical features that can be quickly identified using computer vision models. To further reduce costs associated with building a XyloScope and to take advantage of worldwide near-ubiquity of smartphones, the Forest Products Laboratory also developed the XyloPhone, a 3D printable smartphone attachment. 

Photo Credit

Forest Service photo by Jenna Zukswert

The XyloPhone attachment allows users to take magnified images of wood using a cell phone.

The original XyloPhone design had a price point roughly one twelfth the cost of a XyloScope but permitted nearly identical imaging. Since publication in 2020, researchers have identified cheaper internal components have been found that substantially decrease the price and designed model-specific adapters for dozens of smartphones. To broaden the functionality of the XyloPhone, researchers have developed adapters to hold and center small specimens, to measure slope grain in boards, and to measure slope of grain in cylindrical objects. In addition to using XyloPhones for wood identification, XyloPhones have been used by mycologists, archaeologists, and art researchers for macroscopic investigations in those fields.

While the XyloTron and XyloPhone were originally intended for rapid wood identification, they can also be used to capture magnified images of any other objects, such as fungi, fabric, and leaf surfaces—essentially any material on which they can be placed that will reflect light. The Forest Products Laboratory has developed non-contact versions of both the XyloScope and the XyloPhone for materials too fragile or valuable for direct contact.

Key Findings:

• The XyloTron can rapidly identify wood in the field, which can help detect illegal logging faster than traditional methods of wood identification at ports and other entry points, without need for extensive human expertise.
• The XyloTron system and XyloPhone attachment are open-source and freely available for others to adopt or adapt.
• While the XyloTron and XyloPhone were developed for wood identification, this technology can be used to obtain macroscopic images of any objects of interest.

Approach

Photo Credit

Forest Service photo by Jenna Zukswert

Forest Products Laboratory researcher Alex Wiedenhoeft demonstrates the use of ultraviolet illumination to identify wood samples using the XyloPhone.

The XyloTron uses image-based classification to identify wood samples. Specifically, the XyloScope is used to obtain standardized images of the wood sample, which are then run through a model that has been trained using reference photos obtained from labeled image datasets of known wood species. The hardware and software for the XyloTron is open source; the XyloScope initial design was first published in 2019, and then the XyloTron 2.0 and XyloPhone designs were published in 2020. The updated XyloTron 2.0 design includes ultraviolet illumination, which can help identify wood species that fluoresce, as well as mechanical updates to the XyloScope and open-source software for imaging and wood identification. XyloTron 2.0 can also support the imaging of charcoal for identification.

Wood identification models for the XyloTron have been built and refined for specific regions of the world and for specific types of wood. Research from the Forest Products Laboratory on computer vision wood identification has also evaluated the accuracy of these identification models and refined them over time. Research is currently underway to extract and analyze wood anatomy features from the images that have been taken using the XyloTron, specifically to detect and measure growth ring boundaries and to detect vessels (pores) in hardwoods. 

Outcomes

• Research on computer vision wood identification models has resulted in improved accuracy and refinement for use with the XyloTron.
• XyloTron models have been built and refined for North American commercial hardwoods as well as timbers from Colombia, Ghana, and Peru.