An Object-Based Image Analysis Workflow for Monitoring Shallow-Water Aquatic Vegetation in Multispectral Drone Imagery

Abstract

High-resolution drone aerial surveys combined with object-based image analysis are transforming our capacity to monitor and manage aquatic vegetation in an era of invasive species. To better exploit the potential of these technologies, there is a need to develop more efficient and accessible analysis workflows and focus more efforts on the distinct challenge of mapping submerged vegetation. We present a straightforward workflow developed to monitor emergent and submerged invasive water soldier (Stratiotes aloides) in shallow waters of the Trent-Severn Waterway in Ontario, Canada. The main elements of the workflow are: (1) collection of radiometrically calibrated multispectral imagery including a near-infrared band; (2) multistage segmentation of the imagery involving an initial separation of above-water from submerged features; and (3) automated classification of features with a supervised machine-learning classifier. The approach yielded excellent classification accuracy for emergent features (overall accuracy = 92%; kappa = 88%; water soldier producer’s accuracy = 92%; user’s accuracy = 91%) and good accuracy for submerged features (overall accuracy = 84%; kappa = 75%; water soldier producer’s accuracy = 71%; user’s accuracy = 84%). The workflow employs off-the-shelf graphical software tools requiring no programming or coding, and could therefore be used by anyone with basic GIS and image analysis skills for a potentially wide variety of aquatic vegetation monitoring operations.

Mapping Invasive Lupinus polyphyllus Lindl. in Semi-natural Grasslands Using Object-Based Image Analysis of UAV-borne Images

“…Surprisingly, all four texture attributes (intensity-based and thermal-based) were ineffective in our study. Previous studies (Chabot et al 2018;Silver et al 2019) with OBIA have proven that textural information was useful. However, they utilised many more texture attributes than employed in our study.…”

“…Though it could be shown that the proposed method performed as well as the standard digitising method, it may be criticised that vegetation mapping based on UAV-borne RS data is challenging to scale up (e.g. Chabot et al 2018). In this study, one UAV flight took 20 min (including ground preparation and flight time) to collect data of approximately 0.4 hectares (without thermal sensor).…”

Mapping invasive Lupinus polyphyllus Lindl. in semi-natural grasslands using object-based analysis of UAV-borne images

“…Previous studies (Chabot et al, 2018;Silver et al, 2019) with OBIA have proven that textural information was useful. However, they utilised many more texture attributes than employed in our study.…”

“…Though it could be shown that the proposed method performed as well as the standard digitising method, it may be criticised that vegetation mapping based on UAV-borne RS data is challenging to scale up (e.g. Chabot et al, 2018). In this study, one UAV flight took 20 minutes (including ground preparation and flight time) to collect data of approximately 0.4 hectares (without thermal sensor).…”

Detecting plant species in the field with deep learning and drone technology

“…Some of these challenges can be addressed during the training process using appropriate augmentation strategies, as was shown in this article with regard to brightness augmentation, resulting in a model that offered 27% better F1-score on the test set. However, other variations such as shadow prominence are harder to synthesise using augmentation strategies and as a result must be reflected in the dataset for the model to learn invariance to them.The low sun elevation angle present in the winter months results in visible shadows even at midday, and has previously been noted as unfavourable for data collection byLehmann et al (2017) andChabot et al (2018) who recommended that flights be conducted within two hours of the sun's zenith to achieve the best possible illumination. Variation in annotation confidence in the masks used during training, due to the indistinct edges associated with vegetation, was also shown to affect model performance and was overcome through the use of WMB loss.…”