Optical flow-based branch segmentation for complex orchard environments
This provides a solution for robotics in orchards by reducing the need for large hand-labeled datasets or controlled conditions, though it is incremental as it builds on existing simulation-based methods.
The paper tackles the problem of branch segmentation in visually complex orchard environments by training a neural network solely on simulated RGB and optical flow data, achieving highly accurate and robust performance without real-world training or special equipment.
Machine vision is a critical subsystem for enabling robots to be able to perform a variety of tasks in orchard environments. However, orchards are highly visually complex environments, and computer vision algorithms operating in them must be able to contend with variable lighting conditions and background noise. Past work on enabling deep learning algorithms to operate in these environments has typically required large amounts of hand-labeled data to train a deep neural network or physically controlling the conditions under which the environment is perceived. In this paper, we train a neural network system in simulation only using simulated RGB data and optical flow. This resulting neural network is able to perform foreground segmentation of branches in a busy orchard environment without additional real-world training or using any special setup or equipment beyond a standard camera. Our results show that our system is highly accurate and, when compared to a network using manually labeled RGBD data, achieves significantly more consistent and robust performance across environments that differ from the training set.