DeepForest: Sensing Into Self-Occluding Volumes of Vegetation With Aerial Imaging

2025-02-04

AI Summary

Traditional optical remote sensing struggles to penetrate dense forest canopies and cannot retrieve 3D structural information from self-occluded vegetation. To address this, we propose a multi-focal-plane aerial imaging method leveraging drone-based synthetic aperture dynamic focusing, integrated with a pre-trained 3D convolutional neural network to suppress defocus artifacts and fuse multispectral focal stack data. For the first time, we achieve voxel-level 3D reflectance reconstruction of vegetation beneath closed canopies using only consumer-grade optical cameras—overcoming the long-standing limitation of purely optical approaches in retrieving deep-canopy structural properties. The resulting low-frequency reflectance volume enables quantitative ecological parameter estimation across the full vertical canopy profile (from canopy top to understory), establishing a novel paradigm for立体 (volumetric) monitoring of plant health, growth status, and environmental gradients.

Technology Category

• Computer Vision: 3D Computer Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Access to below-canopy volumetric vegetation data is crucial for understanding ecosystem dynamics. We address the long-standing limitation of remote sensing to penetrate deep into dense canopy layers. LiDAR and radar are currently considered the primary options for measuring 3D vegetation structures, while cameras can only extract the reflectance and depth of top layers. Using conventional, high-resolution aerial images, our approach allows sensing deep into self-occluding vegetation volumes, such as forests. It is similar in spirit to the imaging process of wide-field microscopy, but can handle much larger scales and strong occlusion. We scan focal stacks by synthetic-aperture imaging with drones and reduce out-of-focus signal contributions using pre-trained 3D convolutional neural networks. The resulting volumetric reflectance stacks contain low-frequency representations of the vegetation volume. Combining multiple reflectance stacks from various spectral channels provides insights into plant health, growth, and environmental conditions throughout the entire vegetation volume.

Problem

• Penetrate dense canopy layers
• Measure 3D vegetation structures
• Assess plant health and growth

Innovation

• Synthetic-aperture imaging with drones
• Pre-trained 3D convolutional neural networks
• Volumetric reflectance stacks from spectral channels