DeepCS-TRD, a Deep Learning-based Cross-Section Tree Ring Detector

2025-04-22

AI Summary

This paper addresses the challenging problem of automatic tree-ring detection across multiple tree species (loblolly pine, honey locust, willow) and imaging modalities (microscopy, flatbed scanning, smartphone photography). To overcome limitations of conventional edge-based approaches, we propose the first unified deep learning framework: a U-Net–based end-to-end model augmented with robust multi-resolution and multi-contrast preprocessing and postprocessing strategies. Our key contributions are: (1) the first unified modeling approach enabling cross-species and cross-modality tree-ring detection; (2) the release of two high-quality, expert-annotated benchmark datasets; (3) state-of-the-art performance on macroscopic images of loblolly pine and honey locust, significantly outperforming existing methods; and (4) full open-sourcing of code and an integrated toolchain to advance standardization and reproducibility in dendrochronological analysis.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Here, we propose Deep CS-TRD, a new automatic algorithm for detecting tree rings in whole cross-sections. It substitutes the edge detection step of CS-TRD by a deep-learning-based approach (U-Net), which allows the application of the method to different image domains: microscopy, scanner or smartphone acquired, and species (Pinus taeda, Gleditsia triachantos and Salix glauca). Additionally, we introduce two publicly available datasets of annotated images to the community. The proposed method outperforms state-of-the-art approaches in macro images (Pinus taeda and Gleditsia triacanthos) while showing slightly lower performance in microscopy images of Salix glauca. To our knowledge, this is the first paper that studies automatic tree ring detection for such different species and acquisition conditions. The dataset and source code are available in https://github.com/hmarichal93/deepcstrd

Problem

• Detects tree rings automatically in cross-sections using deep learning
• Works with various image types and multiple tree species
• Improves performance over existing methods in most conditions

Innovation

• Uses U-Net deep learning for edge detection
• Applies to diverse image types and species
• Introduces two public annotated datasets