Tree Trunk Depth Maps and Segmentation Masks for Mobile AR Applications

A dataset from the University of Cambridge contains pairs of depth maps and corresponding tree trunk segmentation masks. Depth maps were captured using a custom Android app with Google ARCore and Huawei's AREngine, with accompanying RGB images of the scenes. The dataset was created by Sofija Stefanović for a machine learning task to improve tree diameter estimation from mobile AR data.

Use Cases

• Training segmentation models to identify tree trunks based on depth maps.
• Improving tree diameter estimation algorithms using depth data from mobile devices.
• Developing and benchmarking multimodal (depth + RGB) computer vision models for environmental sensing.
• Researching the accuracy and utility of consumer-grade AR depth sensors for forestry applications.

Strengths

• Includes paired data of depth maps, segmentation masks, and RGB images, enabling multimodal analysis.
• Data was created for a specific machine learning segmentation task, suggesting a focused application.
• Depth maps were generated using two distinct mobile AR platforms (Google ARCore and Huawei AREngine), allowing for comparison.

Limitations

• Dataset size, row count, and file formats are unknown, which limits suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Cambridge, Sofija Stefanović

Collection Method

Depth maps captured with a custom Android app using Google ARCore and Huawei AREngine, processed with a tree diameter estimation algorithm.