Pseudo-label Generation for Agricultural Robotics Applications

Ciarfuglia, Thomas A.; Motoi, Ionut M.; Saraceni, Leonardo; Nardi, Daniele

doi:10.1109/cvprw56347.2022.00175

Cited by 8 publications

(4 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Addressing works dealing with similar tasks, but on different crops, Ciarfuglia et al [ 19 ] tackled the difficulty of generating pseudo-labels for agricultural robotics applications due to the scarcity of annotated data for fruits in orchards. Thus, they presented a method for producing high-quality pseudo-labels by combining self-supervised and transfer learning.…”

Section: Related Workmentioning

confidence: 99%

Mushroom Detection and Three Dimensional Pose Estimation from Multi-View Point Clouds

Retsinas

Efthymiou

Anagnostopoulou

et al. 2023

Sensors

View full text Add to dashboard Cite

Agricultural robotics is an up and coming field which deals with the development of robotic systems able to tackle a multitude of agricultural tasks efficiently. The case of interest, in this work, is mushroom collection in industrial mushroom farms. Developing such a robot, able to select and out-root a mushroom, requires delicate actions that can only be conducted if a well-performing perception module exists. Specifically, one should accurately detect the 3D pose of a mushroom in order to facilitate the smooth operation of the robotic system. In this work, we develop a vision module for 3D pose estimation of mushrooms from multi-view point clouds using multiple RealSense active–stereo cameras. The main challenge is the lack of annotation data, since 3D annotation is practically infeasible on a large scale. To address this, we developed a novel pipeline for mushroom instance segmentation and template matching, where a 3D model of a mushroom is the only data available. We evaluated, quantitatively, our approach over a synthetic dataset of mushroom scenes, and we, further, validated, qualitatively, the effectiveness of our method over a set of real data, collected by different vision settings.

show abstract

Section: Related Workmentioning

confidence: 99%

Mushroom Detection and Three Dimensional Pose Estimation from Multi-View Point Clouds

Retsinas

Efthymiou

Anagnostopoulou

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…In viticulture [3] Casado-Garcia et al compare various approaches to leverage label information from just 20% of their data and generate pseudo labels for the remaining 80% unlabelled data which they combine to train semi-supervised semantic segmentation. In [6] Ciarfuglia et al use bounding boxes obtained through a grape detector trained on a source dataset to then produce pseudo-labels for other data which are then applied to train an instance segmentation network. To our knowledge, there is no prior work that uses click annotations to obtain pseudo-labels as well as use those for semi-supervised instance segmentation within the agricultural domain.…”

Section: Agricultural Applicationsmentioning

confidence: 99%

Panoptic One-Click Segmentation: Applied to Agricultural Data

Zimmer

Halstead

McCool

2023

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

In weed control, precision agriculture can help to greatly reduce the use of herbicides, resulting in both economical and ecological benefits. A key element here is the ability to locate and segment all the plants (crop & weed) from image data. Modern instance segmentation techniques can achieve this, however, training such systems requires large amounts of handlabelled data which is expensive and laborious to obtain. Weakly supervised training can help to greatly reduce labelling efforts and costs.In this paper we propose panoptic one-click segmentation, an efficient and accurate offline tool to produce pseudo-labels from click inputs and thereby reduce labelling effort when creating novel datasets. Our approach jointly estimates the pixel-wise location of all N objects in the scene, compared to traditional approaches which iterate independently through all N objects. This results in a highly efficient technique with greatly reduced training times. Using just 10% of the data to train our panoptic one-click segmentation approach yields 68.1% and 68.8% mean object intersection over union (IoU) on challenging sugar beet and corn image data respectively, providing comparable performance to traditional one-click approaches while being approximately 12 times (an order of magnitude) faster to train. We demonstrate the practical applicability of our system by generating pseudo-labels from click annotations for the remaining 90% of the data. These pseudo-labels are then used to train Mask R-CNN, in a semisupervised manner, improving the absolute performance (of mean foreground IoU) by 9.4 and 7.9 points for sugar beet and corn data respectively, demonstrating the potential of our approach to rapidly annotate challenging data. Finally, we show that our panoptic one-click segmentation technique is able to recover missed clicks during annotation outlining a further benefit over traditional approaches.

show abstract

“…Some approaches, such as Beck et al (2020), use robotic systems to capture data from a wider range of angles and automatically annotate bounding boxes knowing the location of plants in controlled environments. Other approaches (Fatima and Mahmood, 2021;Ciarfuglia et al, 2022) involve semi-supervised learning, taking advantage of highly precise data to train highperformance models which in turn improve predictions on unlabeled images. Such approaches show potential for not only making new, high-quality agricultural datasets, but even potentially improving upon existing datasets, annotating missed fruits or other objects which may have been missed out on.…”

Section: Annotation Qualitymentioning

confidence: 99%

Standardizing and Centralizing Datasets to Enable Efficient Training of Agricultural Deep Learning Models

Joshi¹,

Guevara²,

Earles³

2022

Preprint

View full text Add to dashboard Cite

In recent years, deep learning models have become the standard for agricultural computer vision. Such models are typically fine-tuned to agricultural tasks using model weights that were originally fit to more general, non-agricultural datasets. This lack of agriculture-specific fine-tuning potentially increases training time and resource use, and decreases model performance, leading an overall decrease in data efficiency. To overcome this limitation, we collect a wide range of existing public datasets for three distinct tasks, standardize them, and construct standard training and evaluation pipelines, providing us with a set of benchmarks and pretrained models. We then conduct a number of experiments using methods which are commonly used in deep learning tasks, but unexplored in their domain-specific applications for agriculture. Our experiments guide us in developing a number of approaches to improve data efficiency when training agricultural deep learning models, without large-scale modifications to existing pipelines. Our results demonstrate that even slight training modifications, such as using agricultural pretrained model weights, or adopting specific spatial augmentations into data processing pipelines, can significantly boost model performance and result in shorter convergence time, saving training resources. Furthermore, we find that even models trained on low-quality annotations can produce comparable levels of performance to their high-quality equivalents, suggesting that datasets with poor annotations can still be used for training, expanding the pool of currently available datasets. Our methods are broadly applicable throughout agricultural deep learning, and present high potential for significant data efficiency improvements.

show abstract

Pseudo-label Generation for Agricultural Robotics Applications

Cited by 8 publications

References 16 publications

Mushroom Detection and Three Dimensional Pose Estimation from Multi-View Point Clouds

Mushroom Detection and Three Dimensional Pose Estimation from Multi-View Point Clouds

Panoptic One-Click Segmentation: Applied to Agricultural Data

Standardizing and Centralizing Datasets to Enable Efficient Training of Agricultural Deep Learning Models

Contact Info

Product

Resources

About