Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Estimating Pasture Biomass from Top-View Images: A Dataset for Precision Agriculture

Qiyu Liao, Dadong Wang, Rebecca Haling, Jiajun Liu, Xun Li, Martyna Plomecka, Andrew Robson, Matthew Pringle, Rhys Pirie, Megan Walker, Joshua Whelan

TL;DR

This work tackles the challenge of accurately estimating pasture biomass to enable precision grazing by introducing a public dataset of 1,162 top-view quadrat images from 19 Australian locations, enriched with NDVI and height measurements and ground-truth biomass broken down into Green, Dead, Clover, Green Dry Matter (GDM), and Dry_Total. It describes a full data pipeline, including geometric normalization, rigorous quality assurance, and a five-target evaluation framework based on a log-transformed weighted $R^2$ metric, with $y_{trans} = \log(1 + y)$. The contribution combines visual, spectral, and structural data on a per-quadrat basis, with training data available alongside richer metadata (date, location, species) and a Kaggle competition to drive AI advances in pasture estimation. By releasing this large, multi-modal, ground-truth dataset, the work enables development of AI-driven tools for improved livestock nutrition, environmental stewardship, and sustainable grazing across diverse temperate pastures.

Abstract

Accurate estimation of pasture biomass is important for decision-making in livestock production systems. Estimates of pasture biomass can be used to manage stocking rates to maximise pasture utilisation, while minimising the risk of overgrazing and promoting overall system health. We present a comprehensive dataset of 1,162 annotated top-view images of pastures collected across 19 locations in Australia. The images were taken across multiple seasons and include a range of temperate pasture species. Each image captures a 70cm * 30cm quadrat and is paired with on-ground measurements including biomass sorted by component (green, dead, and legume fraction), vegetation height, and Normalized Difference Vegetation Index (NDVI) from Active Optical Sensors (AOS). The multidimensional nature of the data, which combines visual, spectral, and structural information, opens up new possibilities for advancing the use of precision grazing management. The dataset is released and hosted in a Kaggle competition that challenges the international Machine Learning community with the task of pasture biomass estimation. The dataset is available on the official Kaggle webpage: https://www.kaggle.com/competitions/csiro-biomass

Estimating Pasture Biomass from Top-View Images: A Dataset for Precision Agriculture

TL;DR

This work tackles the challenge of accurately estimating pasture biomass to enable precision grazing by introducing a public dataset of 1,162 top-view quadrat images from 19 Australian locations, enriched with NDVI and height measurements and ground-truth biomass broken down into Green, Dead, Clover, Green Dry Matter (GDM), and Dry_Total. It describes a full data pipeline, including geometric normalization, rigorous quality assurance, and a five-target evaluation framework based on a log-transformed weighted metric, with . The contribution combines visual, spectral, and structural data on a per-quadrat basis, with training data available alongside richer metadata (date, location, species) and a Kaggle competition to drive AI advances in pasture estimation. By releasing this large, multi-modal, ground-truth dataset, the work enables development of AI-driven tools for improved livestock nutrition, environmental stewardship, and sustainable grazing across diverse temperate pastures.

Abstract

Accurate estimation of pasture biomass is important for decision-making in livestock production systems. Estimates of pasture biomass can be used to manage stocking rates to maximise pasture utilisation, while minimising the risk of overgrazing and promoting overall system health. We present a comprehensive dataset of 1,162 annotated top-view images of pastures collected across 19 locations in Australia. The images were taken across multiple seasons and include a range of temperate pasture species. Each image captures a 70cm * 30cm quadrat and is paired with on-ground measurements including biomass sorted by component (green, dead, and legume fraction), vegetation height, and Normalized Difference Vegetation Index (NDVI) from Active Optical Sensors (AOS). The multidimensional nature of the data, which combines visual, spectral, and structural information, opens up new possibilities for advancing the use of precision grazing management. The dataset is released and hosted in a Kaggle competition that challenges the international Machine Learning community with the task of pasture biomass estimation. The dataset is available on the official Kaggle webpage: https://www.kaggle.com/competitions/csiro-biomass
Paper Structure (16 sections, 2 equations, 2 figures, 2 tables)

This paper contains 16 sections, 2 equations, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Traditional Biomass Measurement Methods
  4. Computer Vision Approaches and Existing Datasets
  5. Data Collection
  6. Study Sites and Sampling Protocol
  7. Laboratory Processing and Quality Control
  8. Image Acquisition Systems
  9. Data Processing and Evaluation
  10. Image Normalization
  11. Data Quality Assurance
  12. Biomass Measurement Standards and Data Integration
  13. Label Structure
  14. Evaluation Protocol
  15. Discussion
  16. ...and 1 more sections

Figures (2)

  • Figure 1: Sample images from the training set and their labels. Images cropped to show the internal section of the 70 x 30cm quadrants, and with dry weight of biomass fractions shown (Green, Dead, Clover).
  • Figure 2: Geographic distribution of sampling sites across Australia trotter2018biomass