BEGIN:VCALENDAR VERSION:2.0 PRODID:-//pretalx//talks.staging.osgeo.org//foss4g-2024-academic-track//spe aker//HWLSJF BEGIN:VTIMEZONE TZID:-03 BEGIN:STANDARD DTSTART:20000101T000000 RRULE:FREQ=YEARLY;BYMONTH=1 TZNAME:-03 TZOFFSETFROM:-0300 TZOFFSETTO:-0300 END:STANDARD END:VTIMEZONE BEGIN:VEVENT UID:pretalx-foss4g-2024-academic-track-JEVCDC@talks.staging.osgeo.org DTSTART;TZID=-03:20241204T164500 DTEND;TZID=-03:20241204T171500 DESCRIPTION:Forest registration is essential for effectively managing natur al resources\, enabling improved tree management (Kattenborn\, Eichel and Fassnacht 2019). This process simplifies urban planning\, allowing for a m ore conscientious approach and significantly contributing to the preservat ion of green areas. The proposal to reduce environmental impact\, survey t ime\, and required effort (Barbosa et al. 2018\; Li et al. 2015\; Beloiu e t al. 2023) has motivated the growing use of computer vision for these tas ks. Today\, this represents a true cartographic revolution. These innovati ons enhance the quality of life in cities by providing accurate and up-to- date data to support critical decisions (Barbosa et al. 2018).\nThis work aims to detect\, classify\, and georeference trees in urban environments u sing image segmentation algorithms applied to aerial and street-level imag es. Several studies use aerial images (Beloiu et al. 2023\; Wäldchen and Mäder 2018\; Mlenek\, Dalla Corte e Santos 2020)\, but our approach seeks to improve the detection and identification of tree species by combining street-level images with aerial images. Our model will be developed with t he algorithms that present the best metrics for species segmentation and c lassification based on related studies. The project also prioritizes using free and open-source software in its development. This not only democrati zes access to robust monitoring and analysis tools but also encourages col laboration and innovation in the geospatial community\, aligning with the values of FOSS4G.\nWe will apply pre-processing techniques to the images t o enhance the model’s accuracy\, including geometric and atmospheric cor rection with QGIS software. Gaussian filters will also be applied to reduc e noise and contrast adjustments to make edges and textures more distinct. After this step\, we will proceed to the feature extraction stage for aut omatic species identification using a machine-learning model. Given the in creasing need for environmental preservation and sustainable management\, identifying and classifying tree species have become solid allies for ecol ogical conservation\, positively impacting urban quality of life.\nTo map the urban area of Rio Paranaíba\, an unmanned aerial vehicle (UAV) drone equipped with a high-resolution camera was used\, capturing images with a 3.5 cm resolution. The UAV was operated autonomously\, flying in parallel strips over the city. A 70% overlap between the images was used\, resultin g in the creation of an accurate orthomosaic of the region\, favoring more accurate georeferencing of the trees. OpenStreetMap software was used to create the orthomosaic. GPS performed georeferencing during the flight. St reet-level images were obtained with a camera that provides 360º coverage . For species classification\, a training dataset was created from samples collected in the field\, both aerial and ground-level. Various machine le arning algorithms\, such as Random Forest\, Support Vector Machine (SVM)\, and Convolutional Neural Networks (CNN)\, were researched and evaluated f or their accuracy in species classification.\nTree identification through images of trunks and leaves presents significant challenges due to high in traclass variability and high interclass similarity. High intraclass varia bility refers to the substantial differences between images of trunks or l eaves of the same tree species caused by lighting variations\, capture ang le\, and tree condition. On the other hand\, high interclass similarity re fers to the very similar visual characteristics between different species\ , making it difficult to distinguish one from another based solely on appe arance. Additionally\, improper color balance adjustments by cameras can i ntroduce unwanted shades\, such as a greenish tint\, further complicating accurate classification. These combined factors make using deep learning f or tree classification a complex and challenging problem (Cotrim et al. 20 19). This technique\, which combines remote sensing with aerial and ground -level images and advanced machine learning techniques\, is expected to pr esent a significant advance in tree species classification. This approach allows for detailed analysis of trunk and leaf textures\, potentially sign ificantly improving species identification accuracy. Studies such as those by Kattenborn\, Eichel and Fassnacht (2019) have demonstrated that CNN-ba sed segmentation (U-net) can achieve an 84% accuracy in vegetation classif ication using high spatial resolution RGB images. The U-net is widely reco gnized for its effectiveness in image segmentation tasks\, especially in h igh-precision and detail scenarios. Its architecture captures complex feat ures\, making it ideal for detecting and classifying specific elements in high-resolution images. Additionally\, the U-net has shown consistent resu lts in various remote sensing applications\, making it a reliable choice f or geospatial data analysis projects. Therefore\, adopting the U-net in th e project can ensure superior tree species identification and mapping perf ormance. This work aligns closely with the themes addressed at the FOSS4G event\, as it demonstrates the practical application of free and open-sour ce software tools in an environmental monitoring context. QGIS\, OpenDrone Map\, and OpenStreetMap exemplify how open technologies can be integrated to solve complex georeferencing and species classification problems. Furth ermore\, the focus on urban areas and the combination of drone and street view data provides valuable insights for the geospatial community\, showin g the feasibility and benefits of free software for urban and environmenta l applications. DTSTAMP:20260514T031456Z LOCATION:Room I SUMMARY:Georeferencing of Urban Trees Using Drones and Ground-Level Imaging \, and Classification of Their Species by Machine Learning - Paulo Roberto Ferreira Maciel\, Rodrigo Smarzaro URL:https://talks.staging.osgeo.org/foss4g-2024-academic-track/talk/JEVCDC/ END:VEVENT END:VCALENDAR