RumexWeeds: A grassland dataset for agricultural robotics

Computer vision can lead toward more sustainable agricultural production by enabling robotic precision agriculture. Vision-equipped robots are being deployed in the fields to take care of crops and control weeds. However, publicly available datasets containing both image data as well from navigational robot sensors scarce. Our real-world dataset RumexWeeds targets detection grassland weeds: Rumex obtusifolius L. crispus includes whole sequences instead individual static images, which is rare for computer datasets, yet crucial applications. It allows robust object detection, incorporating temporal aspects considering different viewpoints same object. Furthermore, additional sensors—GNSS, IMU odometry—which increase robustness, when additionally fed models. In total 5510 images with 15,519 manual bounding box annotations collected at three farms four days summer autumn 2021. Additionally, a subset 340 ground truth pixels-wise annotations. The https://dtu-pas.github.io/RumexWeeds/. this paper we also use provide baseline weed results state-of-the-art detector; way elucidating interesting characteristics dataset.