Over the past decade, Vietnam's wood processing and export industry has experienced strong growth, with export turnover increasing nearly fourfold, reaching US$16.25 billion in 2024 (according to statistics from the General Department of Customs). To meet the large demand for raw materials, Vietnam imports approximately 201 tons of wood annually, with tropical wood accounting for nearly half of the import value – mainly from areas with high risks of illegal logging and unclear origins, such as the Mekong sub-region, the Amazon basin, and Central and West Africa (UNEP-WCMC, 2020). In the past five years, Vietnam has imported more than 1,400 species of wood, including over 60 high-risk species, many of which are listed in the CITES Appendix or easily confused with valuable domestic wood (General Department of Vietnam Customs, 2021-2025). The multi-layered and complex timber supply chain, coupled with misdeclaration of timber species – due to typing errors, lack of knowledge, or deliberate concealment – creates an urgent need for tools to quickly and accurately identify timber species at border crossings.
Given the limited human resources for timber inspection, inconsistent manual procedures, and increasing pressure from regulations on legality and transparency of origin, there is an urgent need for a fast, accurate, and user-friendly timber identification solution – especially for customs and forestry officials working at border crossings and inspection points. This is also a clear requirement and expectation from government agencies in the process of building a more reliable and convenient system to support timber species verification.
Therefore, the project "Supporting the Implementation of the VPA/FLEGT Agreement in Vietnam," authorized by the Federal Ministry of Economic Cooperation and Development of Germany (BMZ), co-implemented by GIZ and the Vietnam Forestry and Forest Protection Department (VNFOREST) with the support of experts from the Research Institute for Forest Industry (RIFI) under the Vietnam Academy of Forest Sciences (VAFS) and AITC Consulting, Technology and Artificial Intelligence Development Company Limited, developed the Vn-WoodID toolkit. In this project, the Research Institute for Forest Industry (RIFI) under the Vietnam Academy of Forest Sciences (VAFS) was responsible for building the macrostructure database, while AITC Consulting, Technology and Artificial Intelligence Development Company Limited developed the AI model and trained it.
Vn-WoodID – is an ecosystem comprising mobile applications available on Google Play and the App Store, and a website. https://woodid.aitc.vn/ and a wood sample library stored at RIFI. The Vn-WoodID application (app.) allows for rapid and highly accurate automatic identification using integrated artificial intelligence (AI) to solve the problem of instant wood species identification on mobile devices, contributing to supporting Vietnam in ensuring the legality, sustainability, and transparency of wood raw materials in the global supply chain.
Why has Vn-WoodID become a breakthrough tool?
- A solid scientific data foundation
The Vn-WoodID database is developed based on:
- The Vn-WoodID database is built from nearly 3,000 wood samples belonging to 260 species, collected in high-risk areas such as Africa, Asia, and South America, including common imported species, newly introduced species, and endangered species often illegally exploited. The database of 260 species includes 188 species imported into Vietnam; 7 species that are rare and valuable timber species listed in Appendices IA and IIA (Decree 84/2021/ND-CP amending and supplementing Decree 06/2019/ND-CP on the management of endangered forest plants and animals and the implementation of the CITES Convention); 12 species whose samples were exchanged with the National Research and Innovation Agency (BRIN) of Indonesia; and 60 species collected (image data) at the Timber Industry Development Department (TIDD) in Takoradi, Ghana.
- After collection, wood samples were identified by two methods: (1) macroscopic and microscopic structural comparison, based on IAWA guidelines, international databases, and control samples; (2) mass spectrometry analysis using DART-TOFMS technology, compared with US-WISC data. Only samples with matching results from both methods were included in the database to ensure the highest accuracy and reliability.
- The automatic identification function of Vn-WoodID is further enhanced by a robust database of macroscopic structural data and detailed images of each wood species, allowing users to directly compare actual wood samples with digitized data. Each species requires a minimum of 300 cross-sectional images – cross-sections perpendicular to the tree trunk – clearly showing growth rings, sapwood/heartwood, early/late wood, wood rays, parenchyma, xylem or tracheids. Images must ensure wood rays are parallel to the image edges, free from cracks, cuts, or light reflections, and should include both wood rays and annual rings.
2. Optimal deep learning technology
Vn-WoodID utilizes the most advanced deep learning model currently available, trained on nearly 100,000 images of wood cross-sections magnified 50 times. The model has been evaluated and achieved a high accuracy of 98.61 TP3T on the test dataset. Furthermore, the model is optimized to run directly on mobile phones, without the need for an internet connection, and with a fast response time of less than 1 second. The application is designed to be minimalist, suitable for non-specialist users; therefore, even customs and forestry officials without in-depth training in wood anatomy can quickly access information.
- Versatile photography capabilities
Vn-WoodID supports direct connection to two groups of lenses: electron microscopes and external magnifying lenses.
- Dino-lite microscope 50x compatible.
- Electron microscope MaxSee cheap
- External camera Attach a macro lens (e.g., Apexel 24x).
- The software interface is user-friendly and easy to operate.
- The process of capturing, storing, and naming image files is standardized according to type identifiers, enabling centralized management and convenient data retrieval.
- Integrating species information databases
Each timber species in the Vn-WoodID library is fully documented with its scientific name, trade name, and natural distribution area. Accompanying information includes density, wood group (hardwood, softwood, etc.), tree morphology, and identifying attributes. Importantly, the data also integrates the species' conservation status, its endangered status according to the CITES list, and current legal regulations related to the management and trade of that timber species. This is a quick search tool to support decision-making regarding the legality and risks of imported timber shipments.
- Support role in inspection and supervision.
Vn-WoodID acts as an initial "filter" to detect anomalies, helping to minimize the time and cost of in-depth examinations. Vn-WoodID is particularly useful for:
- Customs Agency when inspecting and making decisions regarding timber shipments from high-risk areas.
- Forestry officials During inspections and verifications of imported and traded forest products in the market.
- Wood import and processing businesses Fulfilling accountability responsibilities and preventing legal risks due to mistakenly purchasing or incorrectly declaring the species name.
- Teaching and research activities Regarding wood anatomy and biodiversity conservation.
- Future vision
The existing database of wood species images and information serves as the initial foundation for developing the Vn-WoodID wood identification software. However, with the total number of imported species into Vietnam reaching 1,400 and recently 919 (according to Decision No. 2575/QD-BNNMT dated July 7, 2025), expanding the database is essential to enhance application effectiveness.
This expansion can be implemented in an open manner, involving multiple stakeholders:
- RIFI can continue to be the focal point for maintaining and developing the database, provided that resources are secured through public investment, international cooperation, or funded research projects.
- The business community and timber industry associations can participate in providing information on timber species and supplying samples of newly imported species to expand the database.
- Universities and research institutes can integrate this system into their training, practice, and research programs. The physical wood library and digital data are valuable academic resources, supporting learning about species classification, wood structure, image identification, and AI applications in forestry.
Vn-WoodID is not only a software for quickly identifying wood species, but also a digital transformation solution that transparentizes the supply chain, improves management efficiency, and protects forest resources—a testament to the viewpoint of Resolution 57-NQ/TW when identifying science and technology, innovation, and digital transformation as key breakthroughs for developing modern productive forces, improving productivity, efficiency, and sustainability. 2025-07-16_WoodID_GIZ_final
