Publications

Real-Time Joint-Stem Prediction for Agricultural Robots in Grasslands Using Multi-Task Learning

Published in MDPI Agronomy, 2023, 2023

Autonomous weeding robots need to accurately detect the joint stem of grassland weeds in order to control those weeds in an effective and energy-efficient manner. In this work, keypoints on joint stems and bounding boxes around weeds in grasslands are detected jointly using multi-task learning. We compare a two-stage, heatmap-based architecture to a single-stage, regression-based architecture—both based on the popular YOLOv5 object detector. Our results show that introducing joint-stem detection as a second task boosts the individual weed detection performance in both architectures. Furthermore, the single-stage architecture clearly outperforms its competitors with an OKS of 56.3 in joint-stem detection while also achieving real-time performance of 12.2 FPS on Nvidia Jetson NX, suitable for agricultural robots. Finally, we make the newly created joint-stem ground-truth annotations publicly available for the relevant research community.

RumexWeeds: A Grassland Dataset for Agricultural Robotics

Published in Journal of Field Robotics, 2023, 2023

Abstract: Computer vision can lead towards 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 agricultural datasets containing both image data as well as data from navigational robot sensors are scarce. Our real-world dataset RumexWeeds targets the detection of the grassland weeds: Rumex obtusifolius L. and Rumex crispus L.. RumexWeeds includes whole image sequences instead of individual static images, which is rare for computer vision image datasets, yet crucial for robotic applications. It allows for more robust object detection, incorporating temporal aspects and considering different viewpoints of the same object. Furthermore, RumexWeeds includes data from additional navigational robot sensors — GNSS, IMU and odometry — which can increase robustness, when additionally fed to detection models. In total the dataset includes 5,510 images with 15,519 manual bounding box annotations collected at 3 different farms and 4 different days in summer and autumn 2021. Additionally, RumexWeeds includes a subset of 340 ground truth pixels-wise annotations. The dataset is publicly available at https://dtu-pas.github.io/RumexWeeds/. In this paper we also use RumexWeeds to provide baseline weed detection results considering a state-of-the-art object detector; in this way we are elucidating interesting characteristics of the dataset.

Lightweight Monocular Depth Estimation through Guided Decoding

Published in International Conference on Robotics and Automation (ICRA), 2022, 2022

Abstract: We present a lightweight encoder-decoder architecture for monocular depth estimation, specifically designed for embedded platforms. Our main contribution is the Guided Upsampling Block (GUB) for building the decoder of our model. Motivated by the concept of guided image filtering, GUB relies on the image to guide the decoder on upsampling the feature representation and the depth map reconstruction, achieving high resolution results with fine-grained details. Based on multiple GUBs, our model outperforms the related methods on the NYU Depth V2 dataset in terms of accuracy while delivering up to 35.1 fps on the NVIDIA Jetson Nano and up to 144.5 fps on the NVIDIA Xavier NX. Similarly, on the KITTI dataset, inference is possible with up to 23.7 fps on the Jetson Nano and 102.9 fps on the Xavier NX. Our code and models are made publicly available https://github.com/mic-rud/GuidedDecoding .

Self-supervised contrastive learning on agricultural images

Published in Journal: Computers and Electronics in Agriculture, 2021, 2021

Abstract: Agriculture emerges as a prominent application domain for advanced computer vision algorithms. As much as deep learning approaches can help solve problems such as plant detection, they rely on the availability of large amounts of annotated images for training. However, relevant agricultural datasets are scarce and at the same time, generic well-established image datasets such as ImageNet do not necessarily capture the characteristics of agricultural environments. This observation has motivated us to explore the applicability of self-supervised contrastive learning on agricultural images. Our approach considers numerous non-annotated agricultural images, which are easy to obtain, and uses them to pre-train deep neural networks. We then require only a limited number of annotated images to fine-tune those networks in a supervised training manner for relevant downstream tasks, such as plant classification or segmentation. To the best of our knowledge, contrastive self-supervised learning has not been explored before in the area of agricultural images. Our results reveal that it outperforms conventional deep learning approaches in classification downstream tasks, especially for small amounts of available annotated training images where up to 14% increase of average top-1 classification accuracy has been observed. Furthermore, the computational cost for generating data-specific pre-trained weights is fairly low, allowing one to generate easily new pre-trained weights for any custom model architecture or task.

Few-leaf Learning: Weed Segmentation in Grasslands

Published in International Conference on Intelligent Robots and Systems, IROS 2021, 2021

Abstract: Autonomous robotic weeding in grasslands requires robust weed segmentation. Deep learning models can provide solutions to this problem, but they need to be trained on large amounts of images, which in the case of grasslands are notoriously difficult to obtain and manually annotate. In this work we introduce Few-leaf Learning, a concept that facilitates the training of accurate weed segmentation models and can lead to easier generation of weed segmentation datasets with minimal human annotation effort. Our approach builds upon the fact that each plant species within the same field has relatively uniform visual characteristics due to similar environmental influences. Thus, we can train a field-and-day-specific weed segmentation model on synthetic training data stemming from just a handful of annotated weed leaves. We demonstrate the efficacy of our approach for different fields and for two common grassland weeds: Rumex obtusifolius (broad-leaved dock) and Cirsium vulgare (spear thistle). Our code is publicly available at https://github.com/RGring/WeedAnnotator.