Predictive Yield Estimation for Vineyards using Robotic Imaging

Motivation

Accurately forecasting grape yields ia a challenge in viticulture, it directly impacts harvest planning, resource allocation, and financial decision making. Traditional methods often rely on manual sampling, which is time consuming, labour intensive, and provides only a sparse estimate of the entire vineyard’s potential. This can lead to significant inaccuracies in yield prediction.

This project aims to address this challenge by utilising robotics and artificial intelligence. The core objective is to develop a computer vision model that can automatically analyse images of grapes, captured by ground robots to predict grape bunch weights and forecast yields. By training a machine learning model to identify grape bunches, assess their size and density, and correlate these visual features with actual weight data.

This will help vineyard managers with precise and timely insights for optimising harvesting.

• Programming skills in Python.
• An understanding of machine learning, particularly in computer vision (e.g., object detection, image segmentation).
• Experience with deep learning frameworks such as PyTorch or Tensor Flow.
• Ability to work with and process image datasets.
• An understanding of data analysis and model evaluation techniques.
• An analytical mindset with strong problem-solving skills.

This project offers an opportunity to work on precision agriculture and applied AI. You will gain hands‑on experience in developing and deploying machine learning solutions for real-world challenges. This project will enhance your skills in handling complex and variable imaging conditions (e.g., changing light, occlusion of grapes by leaves). This project is suitable as a final year project for students at Lincoln studying Computer Science, Games or Robotics, or as an internship in robotics research.