Conifer Seedling Detection in UAV-Imagery with RGB-Depth Information

TL;DR

This paper enhances conifer seedling detection in UAV imagery by integrating elevation data into the faster-RCNN algorithm, demonstrating that strategic integration of height information significantly improves detection performance.

Contribution

It introduces a novel method of integrating elevation data into object detection networks, specifically optimizing the placement of height information within the architecture.

Findings

Height data integration improves detection accuracy.

Placement of height data after the backbone network is most effective.

Performance gains persist with extended training and higher resolution height data.

Abstract

Monitoring of reforestation is currently being considerably streamlined through the use of drones and image recognition algorithms, which have already proven to be effective on colour imagery. In addition to colour imagery, elevation data is often also available. The primary aim of this work was to improve the performance of the faster-RCNN object detection algorithm by integrating this height information, which showed itself to notably improve performance. Interestingly, the structure of the network played a key role, with direct addition of the height information as a fourth image channel showing no improvement, while integration after the backbone network and before the region proposal network led to marked improvements. This effect persisted with very long training regimes. Increasing the resolution of this height information also showed little effect.