Review of Current Robotic Approaches for Precision Weed Management

Zhang, Wen; Zhang, Miao; Li, Nan; He, Chuangxin; Sun, Teng

doi:10.1007/s43154-022-00086-5

Cited by 58 publications

(40 citation statements)

References 101 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Accurate segmentation of plants in the field enables a range of tasks, from field or crop monitoring [10] through to precise weeding [1]. To achieve this, current state-of-the-art approaches have to be trained in a supervised manner which requires a large number of images to be annotated with a label for each pixel, however, these pixel-wise annotations are expensive and labour intensive to acquire [29]. An alternative is to perform semi-supervised training by using sparse labels for each object [7], [14], we refer to this as weak labelling.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

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

Section: Related Workmentioning

confidence: 99%

“…this domain [11], [29]. Unfortunately, while these techniques can achieve high accuracy in the agricultural domain, they generally require large amounts of labelled data for training.…”

Section: Introductionmentioning

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

show abstract

“…Once identified, weeds that require control can be mechanically or chemically removed. Detection and removal of the targeted weed species may both be performed by a single terrestrial unit (Zhang et al 2022). Alternatively, some authors have suggested combining aerial drones with terrestrial robots.…”

Section: How To Attain Neutral Weed Communitiesmentioning

confidence: 99%

“…Some approaches to species-specific weed management involve sensor-equipped field robots or drones (Zhang et al 2022; Figure 2). The sensors measure features such as weed shape, size, color, texture, and spectral reflectance, then artificial intelligence can be used to identify weed species based on these features (Bawden et al 2017; Pantazi et al 2016; Peteinatos et al 2020; Wang et al 2022).…”

Section: How To Attain Neutral Weed Communitiesmentioning

confidence: 99%

“…Other issues include the difficulty of identifying young seedlings and variable light availability (Zhang et al 2022). Despite these challenges, classification accuracy is improving with the emergence of new machine learning and deep learning strategies, combined with the creation of large training image sets (Zhang et al 2022). For example, Olsen et al (2019) used numerous labeled images of eight Australian weed species and deep learning to achieve a greater than 95% classification accuracy.…”

Section: How To Attain Neutral Weed Communitiesmentioning

confidence: 99%

See 1 more Smart Citation

Neutral weed communities: the intersection between crop productivity, biodiversity, and weed ecosystem services

et al. 2023

View full text Add to dashboard Cite

Weeds are a fundamental component of agroecosystems and, if not appropriately managed, can cause severe crop yield losses. New perspectives on weed management are required because current approaches, such as herbicide application or soil tillage, have significant environmental and agronomic drawbacks. We propose the concept of “neutral weed communities,” which are weed communities that coexist with crops and do not negatively affect their yield and quality compared with weed-free conditions. Management practices that promote neutral weed communities can enable reduced use of herbicides and soil tillage while enhancing ecosystem services and biodiversity. We report scientific evidence of neutral weed communities and survey ecological explanations for why different weed communities have different effects on crop production. We also propose two weed management approaches for obtaining neutral weed communities. The first approach aims to maximize weed biodiversity using traditional approaches such as cropping system diversification and integrated weed management. Higher weed biodiversity is associated with lower dominance of competitive weed species that reduce crop yield. The second approach relies on modern tools such as robots and biotechnology to manipulate the density of specific weed species. This approach can remove highly problematic species and minimize niche overlap between the weeds and crops. Given the complexity of interactions between crops, weeds, and other components of the agroecosystem, we highlight the need for multidisciplinary research to illuminate mechanisms that determine the neutrality of weed communities.

show abstract

Hybridization in agricultural weeds: A review from ecological, evolutionary, and management perspectives

Westbrook,

DiTommaso

2023

American J of Botany

View full text Add to dashboard Cite

Agricultural weeds frequently hybridize with each other or with related crop species. Some hybrid weeds exhibit heterosis (hybrid vigor), which may be stabilized through mechanisms like genome duplication or vegetative reproduction. Even when heterosis is not stabilized, hybridization events diversify weed gene pools and often enable adaptive introgression. Consequently, hybridization may promote weed evolution and exacerbate weed–crop competition. However, hybridization does not always increase weediness. Even when viable and fertile, hybrid weeds sometimes prove unsuccessful in crop fields. This review provides an overview of weed hybridization and its management implications. We describe intrinsic and extrinsic factors that influence hybrid fitness in agroecosystems. We also survey the rapidly growing literature on crop–weed hybridization and the link between hybridization and invasiveness. These topics are increasingly relevant in this era of genetic tools for crop improvement, intensive and simplified cropping systems, and globalized trade. The review concludes with suggested research priorities, including hybridization in the context of climate change, plant–insect interactions, and redesigned weed management programs. From a weed management perspective, hybridization is one of many reasons that researchers and land managers must diversify their weed control toolkits.

show abstract

Review of Current Robotic Approaches for Precision Weed Management

Cited by 58 publications

References 101 publications

Panoptic One-Click Segmentation: Applied to Agricultural Data

Panoptic One-Click Segmentation: Applied to Agricultural Data

Neutral weed communities: the intersection between crop productivity, biodiversity, and weed ecosystem services

Hybridization in agricultural weeds: A review from ecological, evolutionary, and management perspectives

Contact Info

Product

Resources

About