Field-Deployable Computer Vision Wood Identification of Peruvian Timbers.

Area of Science:

• Forestry Science
• Computer Vision
• Artificial Intelligence

Background:

• Illegal logging poses a significant threat to Peruvian, Amazonian, and global tropical forests.
• In Peru, a substantial portion of logging concessions and exported timber are of illegal origin, undermining legal trade and conservation efforts.

Purpose of the Study:

• To develop and evaluate a computer vision system for accurate wood species identification in Peru.
• To combat illegal logging and promote legal, sustainable timber value chains through technology.

Main Methods:

• A convolutional neural network was trained using transfer learning on wood specimen images.
• The XyloTron platform was utilized for image acquisition from six xylaria, covering 228 Peruvian species across 24 classes.
• The system was evaluated on independent wood specimens not used in training.

Main Results:

• The trained models achieved 97% accuracy in five-fold cross-validation.
• Top-1 and top-2 classification accuracies on independent specimens were 86.5% and 92.4%, respectively.
• This represents the first national-scale computer vision wood identification system evaluated on independent scientific specimens across multiple sites and users.

Conclusions:

• The developed AI system demonstrates high accuracy and readiness for real-world field deployment.
• This technology offers an affordable, scalable solution for monitoring, incentivizing, and monetizing legal and sustainable wood value chains.
• The system provides a crucial tool for combating illegal logging and supporting legitimate timber industries.