From Orthomosaics to Raw UAV Imagery: Enhancing Palm Detection and Crown-Center Localization

2025-09-15

AI Summary

Existing palm tree detection and crown-center localization methods in tropical forests suffer from low spatial accuracy, primarily due to artifacts and geometric distortions introduced during orthomosaic generation. Method: This study proposes an end-to-end detection framework operating directly on raw UAV imagery—bypassing orthorectification—while introducing crown-center keypoints as auxiliary supervision to jointly optimize object detection and precise localization. We conduct the first systematic comparison of raw versus orthorectified imagery under both in-domain and cross-domain settings. Contribution/Results: An enhanced multi-task detection architecture maintains high recall while significantly improving single-tree geolocation accuracy—reducing mean localization offset by 32.7%. Raw imagery proves more robust for field deployment, and integrating crown-center annotations enables centimeter-level geographic coordinate estimation for individual trees, thereby supporting fine-grained biodiversity monitoring and precision forest management.

Technology Category

• Computer Vision: Object Detection & Categorization
• Intelligent Robots: Localization, Mapping, and Navigation
• Machine Learning: Ensemble Methods

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Accurate mapping of individual trees is essential for ecological monitoring and forest management. Orthomosaic imagery from unmanned aerial vehicles (UAVs) is widely used, but stitching artifacts and heavy preprocessing limit its suitability for field deployment. This study explores the use of raw UAV imagery for palm detection and crown-center localization in tropical forests. Two research questions are addressed: (1) how detection performance varies across orthomosaic and raw imagery, including within-domain and cross-domain transfer, and (2) to what extent crown-center annotations improve localization accuracy beyond bounding-box centroids. Using state-of-the-art detectors and keypoint models, we show that raw imagery yields superior performance in deployment-relevant scenarios, while orthomosaics retain value for robust cross-domain generalization. Incorporating crown-center annotations in training further improves localization and provides precise tree positions for downstream ecological analyses. These findings offer practical guidance for UAV-based biodiversity and conservation monitoring.

Problem

• Detecting palm trees in raw UAV imagery versus orthomosaics
• Evaluating crown-center annotations for improved tree localization
• Assessing cross-domain transfer performance for forest monitoring

Innovation

• Raw UAV imagery enhances palm detection
• Crown-center annotations improve localization accuracy
• Keypoint models provide precise tree positions