THRUST LOG-IQ

THRUST LOG-IQ: Image-based Quantification of Logs

Problem: Access to forests is crucial for various reasons, ranging from environmental conservation to economic utilization and recreational purposes. Maintaining clear roads and paths within forests is essential to facilitate sustainable forest management practices, allowing foresters, researchers, and conservationists to monitor and protect forest ecosystems effectively. Accessible forests enable timely intervention in case of natural disturbances such as wildfires, insect infestations, or disease outbreaks, thus minimizing potential damage and promoting forest health. Furthermore, clear access routes facilitate responsible logging operations, ensuring the sustainable harvesting of timber resources while minimizing environmental impact. Additionally, accessible forests provide opportunities for recreational activities such as hiking, camping, and wildlife observation, fostering public appreciation for nature and supporting local tourism economies. Therefore, blocking forest access with fallen trees hinders not only the effective management and utilization of forest resources but also restricts the enjoyment and appreciation of these valuable natural landscapes by the wider community.

Purpose: Timely identification of obstructed access to forests is paramount for the sustainable forest management. Our “THRUST LOG-IQ: Image-based Quantification of Logs” solution leverages UAV (Unmanned Aerial Vehicle) inspections and AI-based analytics to identify and quantify fallen trees within forested areas. Our technological approach relies on our in-house developed cutting-edge GreenBee UAV with our custom integrated AI-based automated analytics software for deadwood identification and classification. As a deep-tech start-up working across the value chain, we are developing both the hardware and the software of the THRUST LOG-IQ system. Though the highest value-added is achieved when using THRUST LOG-IQ software together with the high-capacity long-range THRUST GreenBee UAV, the developed software bundle is compatible with any standard UAV data. This way, THRUST LOG-IQ will smoothly integrate into the Drone Innovation Platform (DIP) and its functionality will complement the versatile CHAMELEON ecosystem.


Strengths: A distinctive aspect of our methodology is the end-to-end control over data. As introduced before, we employ our in-house UAV fleet for data collection, providing us with a means to acquire high-quality, geospatially referenced data in a variety of application domains. Furthermore, data annotation, a critical component of the machine learning pipeline, is carried out in-house as well, enabling us to curate datasets that are tailored to the unique needs of each project. Instant communication among the in-house data collection, data processing, model training, and software development teams enables speedy and smooth delivery of customized analytics solutions. Our commitment to technical excellence is fortified by our ongoing collaboration with forestry experts. These experts play a pivotal role in shaping the trajectory of our data analysis and AI model training processes, ensuring their alignment with the specific requirements of forestry applications.

Ethics: We organize all our processes so that they are not only technically robust but also aligned with key ethical and regulatory standards. We are committed to the principles of Trustworthy AI, ensuring that our solutions are transparent, accountable, and ethically sound throughout the development and deployment process.

Vision: During this THRUST LOG-IQ project have capitalized our superior data gathering capabilities and created a large real-life, field-gathered database of deadwood pictures, which we have used to train our AI/DL based software to beyond state-of-the-art functionality: THRUST LOG-IQ vision is to develop software that would be able to identify large trees fallen on forest roads to evaluate forest accessibility after extreme weather events and then expand this functionality for the software to be able to assess wind damage in the whole forest area and account as well as classify deadwood for biodiversity conservation. This product functionality is relying on aforementioned GreenBee UAV capability to carry multiple high-resolution sensors and to inspect hundreds of kilometres of forest roads (or hundreds of hectares of forests) at a time.


Functionality: “THRUST LOG-IQ: Image-based Quantification of Logs” solution takes geotagged aerial UAV imagery as an input and utilizes custom trained DL models combined with GIS processing to detect, characterize, locate, and report windthrows. The analytics flow consist of 3 key steps: 1) detection of individual wind-felled trees, 2) geolocation, and 3) reporting. Our AI analytics solutions are grounded in the application of cutting-edge AI models, utilizing the PyTorch framework, renowned for its flexibility and scalability in the development of custom deep learning-based analytics solutions. In the domain of object detection, our solutions are underpinned by YOLOv8, state-of-the-art architectures that excel in high-precision object identification within complex scenarios. Detection models have been trained using our own meticulously annotated datasets, based on the aerial imagery collected during UAV flights using our THRUST GREENBEE drones. Each detected windthrow instance is assigned GIS metadata for geolocation and the final results are be reported as a GIS vector layer with attribute tables.


Service modes: Full THRUST LOG-IQ services can be utilized in two modes:
          1) together with aerial inspection using THRUST UAVs and
          2) as large-scale aerial imagery analytics services for forest diagnostics.


Support: The THRUST LOG-IQ project has indirectly received funding from the European Union’s Horizon Europe research and innovation action programme, via the CHAMELEON Open Call 1 issued and executed under the CHAMELEON project (Grant Agreement no. 101060529).