Wood Identification Mobile Phone Application Developer

Organizational Setting:

For more than four decades, FAO has been working with the Government of the Philippines, civil society, community-based organizations and the private sector to address challenges in the agriculture, fisheries and forestry sector. Joint efforts have included increasing sustainability in agricultural production, promoting value-adding practices, improving post-harvest management, enhancing productivity and increasing the resilience of agriculture-based livelihoods to natural disasters, climatic hazards and armed conflict.

Reporting Lines:

The Wood Identification Mobile Phone Application Developer will be under the overall guidance of the FAO Representative in the Philippines ad interim, the direct supervision of the Assistant FAO Representative for Programme. He/she will also work under the technical supervision of the FLEGT Forestry Officer based in FAO Regional Office for Asia and the Pacific.

Technical Focus:

In late 2016, the FAO-EU FLEGT Programme Engagement in the Philippines started its support to improve forest governance, reduction of illegal logging and promotion of trade of legally harvested forest products. In 2017, the Roadmap of the FAO-EU FLEGT Programme Engagement in the Philippines was developed. Among the thematic areas to be addressed in the Roadmap is the development or strengthening of the country’s national verification procedures and mechanisms. The Forest Management Bureau (FMB) identified the need for a mechanism to ensure that all timber and lumber products traded on the domestic market are properly and correctly identified. It was proposed to develop IT-based solutions for wood species identification that will hasten accurate and quick identification of all timber and lumber products while in transit and which are subjected to random checks and inspection. The ultimate objective is to rollout a Wood ID application for mobile android phones that can be deployed in the field for use by field officers of the Department of Environment and Natural Resources (DENR). The Philippines’ National Forest Stock Monitoring System (NFSMS) will benefit from and be complemented by this quick, automated and accurate wood identification during inspection of timber or logs and lumber forest being transported as part of NFSMS law enforcement function.

Currently, the Forest Products Research and Development Institute (FPRDI) under the Department of Science and Technology (DoST), has taken initial efforts to promote wood identification in the country. The FPRDI is the authority in wood identification and timber products in the Philippines and has published numerous books on wood identification. FPRDI has a Wood and Herbarium Library (i.e. Xylarium) which is an internationally recognized facility housing 16,348 specimens of Philippine and Foreign wood and 2,631 preserved plant specimens. Relative to this, FPRDI is exploring the use of machine vision-based automated wood identification systems. Through this project, the Forest Products Research and Development Institute (FPRDI) and technical experts from the Center for Wood Anatomy Research (CWAR) of the USFS will train an image-based Wood Identification model of approximately 30 initial woods, using machine learning software. The XyloTron platform will be used, given that it is open-sourced and adaptable; XyloTron hardware and software requirements can be found here and here.

The Forest Management Bureau (FMB) of the Department of Environmental and Natural Resources (DENR) and DENR Field Offices nationwide, are responsible for enforcement of forestry laws, rules, and regulations pursuant to Executive Order 192. FMB-DENR staff will be the primary users of the proposed Wood Identification Mobile Phone Application, and FMB will provide oversight over the development of the proposed Wood Identification Mobile Phone Application Software using the XyloPhone peripheral for on-device wood identification.

Tasks and responsibilities

The Wood Identification Mobile Phone Application Developer will develop a wood identification mobile application which can implement the XyloTron-based wood identification model developed by FPRDI and CWAR for selected endemic, native and exotic timber species. The application should be able to deliver quick, and reliable results on mobile phones using the application and the XyloPhone peripheral (the open-sourced hardware designs for which are available here), and be ready for deployment by DENR field personnel.

The consultant will undertake the following specific tasks:

1. Develop mobile phone application software for wood identification

• The consultant will develop a mobile application that can use the trained XyloTron model developed by FPRDI and partners to implement on-device inference to perform image-based wood identification using images captured with the XyloPhone peripheral. The application would use On-Device-Inference, to ensure wood identification does not require web or network connectivity which can’t be guaranteed when using the application in field conditions. While it is anticipated that the mobile application will be used on Android Smartphones, given that Android phones are more commonly used in the Philippines, ideally the application software will be platform-agnostic to allow for cross-platform usage. The application software should have a graphical user interface (GUI) to take the macro photo of a wood’s cross-section with accompanying XyloPhone imaging peripheral. It should be designed as “proof of concept” such that the taxa included in the model can be modified based on field evaluation metrics.
• The consultant should collaborate with FPRDI and project partners as needed during the XyloTron model training phase, to ensure understanding of the technology and processes that may affect the final design and functionality of the mobile application software.

2. Conduct lab and/or field test of the wood identification application

• The consultant will, in close contact with technical experts at FPRDI (and CWAR as needed) conduct lab and field testing to validate the wood identification application software, in close collaboration with FPRDI, FMB and following the technical advice provided by project partners. They will conduct a User Acceptance Test (UAT) of the mobile application in both the lab and the field, to ascertain whether the mobile advice can indeed quickly and reliably identify the wood species. In the event that domestic and international travel is still restricted at the time for testing, FPRDI, FMB and/or project partners will facilitate access to a secondary digital wood collection of the same species, which can be used as a proxy to test the model and application function.
• During the test, the mobile application on the phone will be used to take photos of woods included in the identification model. The test will be deemed successful if the forensically validated field performance of the mobile phone, on-device model performs within 20 percentage points of the in silico cross-validation results of the parent XyloTron model. Any drop in accuracy less than 20 points will be a strong sign of success, and will inform the sampling and use manual developed as a deliverable, below.
• Based on the results of the UAT, the IT Expert will make necessary updates to the application software before submitting the final source code to FMB and FAO.

CANDIDATES WILL BE ASSESSED AGAINST THE FOLLOWING

Minimum Requirements

• Advanced university degree in computer science, software engineering, data science, or a related field

• At least ten (10) years of relevant experience in mobile phone application development that includes demonstration of skills in cross-platform development, UX/UI design, usage of graphical user interfaces (GUI)

• A national of the Philippines

FAO Core Competencies

• Results Focus
• Teamwork
• Communication
• Building Effective Relationships
• Knowledge Sharing and Continuous Improvement

Technical/Functional Skills

• Track record of success in developing user-friendly mobile phone applications for clients
• Previous experience in deploying trained machine learning models for on-device inference
• General knowledge and skills in the use of standard office software, such as MS Office (Word, Excel, Power Point).
• Language requirement: Excellent written and oral communication skills in English and Tagalog