Thumbnail Image

Precision monitoring of leaf-cutting ant nests in sub-orbital RGB images using deep learning

XV World Forestry Congress, 2-6 May 2022

Year of publication 2022

Place of publication Rome, Italy ;

Pages 1p.

Author Santos, A., Gonçalves Biesseck, B. J., Lima Santos, I. C., et al.

Publisher FAO ;

Product type Document

Synopsis (short abstract) Leaf-cutting ants are the main pests of plantation forests in South America, causing severe defoliation, leading to production losses, plant mortality and increased susceptibility to other insects. Chemical control with sulfluramid active ingredient is the most used method. This proposal aimed to develop an innovative method for leaf-cutting ants nest detection in sub-orbital RGB images using deep learning techniques. The study was carried out in a 6-month-old eucalyptus stand with 91.3 hectares in the municipality of Três Lagoas, in the state of Mato Grosso do Sul, Brazil. The images of the stand were collected by a DJI Phantom 4 Advanced aircraft with an RGB camera and processed to produce an orthomosaic with a ground-level resolution of 5.2 cm/pix. The final orthomosaic was cropped in sub-images of 98 x 81 pixels. Sub-images that contained ant nests were labelled using bounding boxes. The database used in experiments consists of 2465 images containing leaf-cutting ant nests and 2465 images of ants' nest absence (background). The detection algorithm used as the deep learning framework based on the YOLO convolutional neural network architecture. The quality of its predictions was evaluated by accuracy, Kappa, sensitivity, specificity, and absolute mean error (MAE) metrics between training and validation samples. YOLO achieved 98.45% accuracy and YOLO 0.49% MAE as the best performances in nests measuring task, demonstrating the high complexity of detecting this target type. Obtained results show that YOLO is a promising approach for precision monitoring of leaf-cutting ant nests in sub-orbital RGB images and can contribute to reduce and optimize insecticide use in plantation forests, which is aligned with the UN Sustainable Development Goals (SDGs), consisting of responsible consumption and production (Goal 12), and terrestrial life (Goal 15). Keywords: Monitoring and data collection, Innovation, Knowledge management ID: 3622337

Statutory body / Meeting title World Forestry Congress (WFC)

Language English

Also available in:
No results found.

Related items

Showing items related by metadata.

  • Thumbnail Image
    Document
    Timber loading productivity of remote controlled forestry equipment through image of monitor
    XV World Forestry Congress, 2-6 May 2022
    2022
    Also available in:
    No results found.
    Forest operations like timber harvesting have already been mechanized to reduce hazards to the worker and increase productivity. However, timber harvesting operations have still been considered potentially dangerous and expensive on steep terrain. Teleoperation, to control the timber harvesting machine at a distance, has the potential to improve the safety, productivity and efficiency of harvesting operations on steep terrain. To verify the effects of teleoperation, an experimental prototype system of a monitor image-based teleoperation was constructed using a real forestry machine. In this study, the productivity of excavator based grapple loader, which is one of the most used mechanized harvesting equipment in the timber production, was analyzed using time-study method. Factors like skill and age of operators, influencing loader productivity in timber loading operation were also evaluated by statistical analysis. Productivity analysis results showed that less experienced operators were more productive than experienced operators for teleoperation through image of monitors in the operator cabin. These results are shown to be unfamiliar to the monitor image and different loading operation pattern by operators. According to the results, the monitor image-based teleoperation system of forestry machine need to improve the resolution and installation position of camera. It was expected that additional studies will be needed for real-time remote control of forestry machine in the future. Keywords: Research, Financial mechanisms ID: 3622426
  • Thumbnail Image
    Document
    LIDAR image–based fuel construction in a computational fluid dynamics simulation domain
    XV World Forestry Congress, 2-6 May 2022
    2022
    Also available in:
    No results found.
    LiDAR image-based vegetation fuel construction in a computational fluid dynamic (CFD) simulation domain was investigated. Using LiDAR images to convey fuel information to CFD would improve the accuracy of wildfire spread prediction. The obtained vegetation information using LiDAR appears as point signals in LiDAR images, and the point signals were dispatched to nodes using the K-D tree algorithm. Then, each node is transferred to the meshing algorithm along with the number of signals and location information. In a CFD domain, 3-dimension vegetation fuel information is reconstructed, and fuel mass is estimated by using the number of signals within each mesh. It appears that utilizing LiDAR images to obtain fuel information improves the accuracy in fuel shapes and mass distribution compared to the conventional way that assigns pre-determined shape and mass distribution for each vegetation. It is expected that the outcomes of this research would improve the liability of CFD-based wildfire prediction. Keywords: Sustainable forest management, Research, Climate change ID: 3617419
  • Thumbnail Image
    Document
    Analyzing the occurrence trend of sediment-related disasters and post-disaster recovery cases in mountain regions in North Korea based on a literature review and satellite image observations
    XV World Forestry Congress, 2-6 May 2022
    2022
    Also available in:
    No results found.
    This study investigated spatiotemporal trends of sediment-related disasters in North Korea from 1960 to 2019 and post-disaster recovery cases based on a literature review and satellite images. Results showed that occurrence status of sediment-related disasters was initially externally reported in 1995 (during the Kim Jongil era); their main triggering factor was heavy summer rainfall. Furthermore, forest degradation rate was positively correlated with population density (R2 = 0.4347, p = 0.02) and occurrence number of sediment-related disasters was relatively high on the west coast region, where both variables showed high values. This indicates that human activity was a major cause of forest degradation and thus, significantly affected sediment-related disasters in mountain regions. Finally, sediment- related disasters due to shallow landslides, debris flow, and slow-moving landslides were observed in undisturbed forest regions and human-impacted forest regions, including terraced fields, opencast mines, forest roads, and post-wildfire areas, via satellite image analysis. These disaster-hit areas remained mostly abandoned without any recovery works, whereas hillside erosion control work (e.g., treeplanting with terracing) or torrent erosion control work (e.g., check dam, debris flow guide bank) were implemented in certain areas. These findings can provide reference information to expand inter-Korean exchange and cooperation in forest rehabilitation and erosion control works of North Korea. Keywords: Climate change, Deforestation and forest degradation, Sustainable forest management, Monitoring and data collection, Research ID: 3616353
Show more

Users also downloaded

Showing related downloaded files

No results found.