2018
DOI: 10.1109/lra.2017.2774979
|View full text |Cite
|
Sign up to set email alerts
|

weedNet: Dense Semantic Weed Classification Using Multispectral Images and MAV for Smart Farming

Abstract: Abstract-Selective weed treatment is a critical step in autonomous crop management as related to crop health and yield. However, a key challenge is reliable, and accurate weed detection to minimize damage to surrounding plants. In this paper, we present an approach for dense semantic weed classification with multispectral images collected by a micro aerial vehicle (MAV). We use the recently developed encoder-decoder cascaded Convolutional Neural Network (CNN), Segnet, that infers dense semantic classes while a… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

4
173
0
1

Year Published

2019
2019
2021
2021

Publication Types

Select...
5

Relationship

0
5

Authors

Journals

citations
Cited by 283 publications
(178 citation statements)
references
References 39 publications
4
173
0
1
Order By: Relevance
“…However, since our domain is characterized by severely limited data set sizes, only three classes, and a high operational speed requirement, it was deemed necessary to simplify the aforementioned network from 13 layers to 4 to both decrease the number of involved parameters and the duration of training and testing. Our decision is also supported by the similar observations made in past papers (Lottes et al, ; Milioto et al, ; Sa et al, ) about how relatively simple networks are sufficient for dealing with this problem.…”
Section: Approachsupporting
confidence: 81%
See 3 more Smart Citations
“…However, since our domain is characterized by severely limited data set sizes, only three classes, and a high operational speed requirement, it was deemed necessary to simplify the aforementioned network from 13 layers to 4 to both decrease the number of involved parameters and the duration of training and testing. Our decision is also supported by the similar observations made in past papers (Lottes et al, ; Milioto et al, ; Sa et al, ) about how relatively simple networks are sufficient for dealing with this problem.…”
Section: Approachsupporting
confidence: 81%
“…More recently, real‐time systems to distinguish crops from weeds have been developed (Lottes et al, ; Milioto et al, ; Sa et al, ) based on SegNet (Badrinarayanan, Kendall, & Cipolla, ), a deep encoder–decoder convolutional network architecture for multiclass pixel‐wise segmentation. The performance of the classifiers trained on different combinations of input channels as well as on NDVI images calculated from the Red and NIR channels is compared in Sa et al (), without a significant difference between the different combinations. They also studied the impact of using a pretrained (on generic color images) network model followed by fine‐tuning; however, they concluded that this did not have a significant impact on the output.…”
Section: Related Workmentioning
confidence: 99%
See 2 more Smart Citations
“…Classification and regression trees were used to perform semantic segmentation on data collected by a six‐band multispectral system in Bac, Hemming, & vanHenten, . With the advent of deep‐learning, many methods for semantic segmentation based on convolutional neural networks (CNNs) have been proposed for color imagery (Milioto, Lottes, & Stachniss, ) and multispectral data (Lottes, Behley, Milioto, & Stachniss, ; McCool, Perez, & Upcroft, ; Sa et al, ).…”
Section: Resultsmentioning
confidence: 99%