Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Infield Navigation: leveraging simulated data for crop row detection

Rajitha de Silva, Grzegorz Cielniak, Junfeng Gao

TL;DR

This work introduces an automated pipeline to generate labelled images for crop row detection in simulation domain and suggests the utilization of small real-world datasets along with additional data generated by simulations to yield similar crop Row detection performance as that of a model trained with a large real world dataset.

Abstract

Agricultural datasets for crop row detection are often bound by their limited number of images. This restricts the researchers from developing deep learning based models for precision agricultural tasks involving crop row detection. We suggest the utilization of small real-world datasets along with additional data generated by simulations to yield similar crop row detection performance as that of a model trained with a large real world dataset. Our method could reach the performance of a deep learning based crop row detection model trained with real-world data by using 60% less labelled real-world data. Our model performed well against field variations such as shadows, sunlight and grow stages. We introduce an automated pipeline to generate labelled images for crop row detection in simulation domain. An extensive comparison is done to analyze the contribution of simulated data towards reaching robust crop row detection in various real-world field scenarios.

Towards Infield Navigation: leveraging simulated data for crop row detection

TL;DR

This work introduces an automated pipeline to generate labelled images for crop row detection in simulation domain and suggests the utilization of small real-world datasets along with additional data generated by simulations to yield similar crop Row detection performance as that of a model trained with a large real world dataset.

Abstract

Agricultural datasets for crop row detection are often bound by their limited number of images. This restricts the researchers from developing deep learning based models for precision agricultural tasks involving crop row detection. We suggest the utilization of small real-world datasets along with additional data generated by simulations to yield similar crop row detection performance as that of a model trained with a large real world dataset. Our method could reach the performance of a deep learning based crop row detection model trained with real-world data by using 60% less labelled real-world data. Our model performed well against field variations such as shadows, sunlight and grow stages. We introduce an automated pipeline to generate labelled images for crop row detection in simulation domain. An extensive comparison is done to analyze the contribution of simulated data towards reaching robust crop row detection in various real-world field scenarios.
Paper Structure (12 sections, 2 equations, 10 figures, 4 tables)

This paper contains 12 sections, 2 equations, 10 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset
  4. Data Categories
  5. Methodology
  6. Simulated Data Generation
  7. Training with Real World Data
  8. Training with Simulated Data
  9. Experimental Evaluation
  10. IoU Analysis
  11. Categorical Analysis
  12. Conclusion

Figures (10)

  • Figure 1: Sample image and respective ground truth label mask
  • Figure 2: Husky Robot with Realsense Cameras
  • Figure 3: Sample images from data categories in real-world dataset
  • Figure 4: Sugar Beet Field Simulation with Husky Robot
  • Figure 5: Sample Image from Simulated Dataset (Left) and Sample Image from Real World Dataset (Right)
  • ...and 5 more figures