Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Vision-based automatic fruit counting with UAV

Hubert Szolc, Mateusz Wasala, Remigiusz Mietla, Kacper Iwicki, Tomasz Kryjak

TL;DR

The paper addresses automatic fruit counting from UAVs in smart agriculture by combining a HSV-based, RGB-D vision pipeline with depth-informed validation and a trajectory-planning layer. It emphasizes fast, on-board processing using classical image techniques and a Fast-TSP–TOPP-RA planning loop to minimize flight time while ensuring safety. In simulation on 500 randomized missions, the approach achieves strong fruit-count accuracy and low collision rates, and it secured 6th place (84.83/100) in the ICUAS 2024 UAV competition finals, demonstrating competitiveness under realistic constraints. The work offers a practical, explainable solution suitable for real-time deployment and outlines future steps toward real-world testing and heterogeneous hardware deployment.

Abstract

The use of unmanned aerial vehicles (UAVs) for smart agriculture is becoming increasingly popular. This is evidenced by recent scientific works, as well as the various competitions organised on this topic. Therefore, in this work we present a system for automatic fruit counting using UAVs. To detect them, our solution uses a vision algorithm that processes streams from an RGB camera and a depth sensor using classical image operations. Our system also allows the planning and execution of flight trajectories, taking into account the minimisation of flight time and distance covered. We tested the proposed solution in simulation and obtained an average score of 87.27/100 points from a total of 500 missions. We also submitted it to the UAV Competition organised as part of the ICUAS 2024 conference, where we achieved an average score of 84.83/100 points, placing 6th in a field of 23 teams and advancing to the finals.

Vision-based automatic fruit counting with UAV

TL;DR

The paper addresses automatic fruit counting from UAVs in smart agriculture by combining a HSV-based, RGB-D vision pipeline with depth-informed validation and a trajectory-planning layer. It emphasizes fast, on-board processing using classical image techniques and a Fast-TSP–TOPP-RA planning loop to minimize flight time while ensuring safety. In simulation on 500 randomized missions, the approach achieves strong fruit-count accuracy and low collision rates, and it secured 6th place (84.83/100) in the ICUAS 2024 UAV competition finals, demonstrating competitiveness under realistic constraints. The work offers a practical, explainable solution suitable for real-time deployment and outlines future steps toward real-world testing and heterogeneous hardware deployment.

Abstract

The use of unmanned aerial vehicles (UAVs) for smart agriculture is becoming increasingly popular. This is evidenced by recent scientific works, as well as the various competitions organised on this topic. Therefore, in this work we present a system for automatic fruit counting using UAVs. To detect them, our solution uses a vision algorithm that processes streams from an RGB camera and a depth sensor using classical image operations. Our system also allows the planning and execution of flight trajectories, taking into account the minimisation of flight time and distance covered. We tested the proposed solution in simulation and obtained an average score of 87.27/100 points from a total of 500 missions. We also submitted it to the UAV Competition organised as part of the ICUAS 2024 conference, where we achieved an average score of 84.83/100 points, placing 6th in a field of 23 teams and advancing to the finals.

Paper Structure

This paper contains 7 sections, 1 equation, 4 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Proposed solution
  4. Results and discussion
  5. Evaluation metrics
  6. Evaluation results
  7. Summary

Figures (4)

  • Figure 1: The overview of our vision-based system for automatic fruit counting with UAV. Proposed solution uses a streams from an RGB camera and a depth sensor and process them with classical image processing algorithms. Additionally, our system enables planning and execution of flight trajectories, optimizing for minimal flight time and distance covered.
  • Figure 2: Example of scene configuration in Gazebo simulation.
  • Figure 3: Scheme of our algorithm for fruits detection in simulation environment.
  • Figure 4: Test missions parameters distribution for 500 total number of cases.