A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
DeepVerge: Classification of roadside verge biodiversity and conservation potential
Abstract Grasslands are increasingly modified by anthropogenic activities, and species-rich grasslands have become rare habitats in the UK. However, grassy roadside verges often contain conservation priority plant species that should be targeted for protection. Identification of verges with high conservation potential represents a considerable challenge for ecologists, driving the development of automated methods to make up for the shortfall of relevant expertise nationally. Using survey data from 3900 km of roadside verges alongside publicly available street-view imagery, we present DeepVerge: a deep learning-based method that can automatically survey sections of roadside verge based on the presence of positive indicator species. Using images and ground truth survey data from the rural UK county of Lincolnshire, DeepVerge achieved a mean accuracy of 88% and a mean F 1 score of 0.82 based on a five-fold cross-validation. We argue that with suitable locale-specific fine-tuning, such a method may be used by local authorities to identify new local wildlife sites, and aid management and environmental planning in line with legal and government policy obligations.
Highlights A system for remote surveying of UK roadside verges was developed. Street-view imagery was used to train a deep learning classifier. The model can identify roadside verges with high conservation potential. The automated system reduces the need for manual field surveys.
DeepVerge: Classification of roadside verge biodiversity and conservation potential
Abstract Grasslands are increasingly modified by anthropogenic activities, and species-rich grasslands have become rare habitats in the UK. However, grassy roadside verges often contain conservation priority plant species that should be targeted for protection. Identification of verges with high conservation potential represents a considerable challenge for ecologists, driving the development of automated methods to make up for the shortfall of relevant expertise nationally. Using survey data from 3900 km of roadside verges alongside publicly available street-view imagery, we present DeepVerge: a deep learning-based method that can automatically survey sections of roadside verge based on the presence of positive indicator species. Using images and ground truth survey data from the rural UK county of Lincolnshire, DeepVerge achieved a mean accuracy of 88% and a mean F 1 score of 0.82 based on a five-fold cross-validation. We argue that with suitable locale-specific fine-tuning, such a method may be used by local authorities to identify new local wildlife sites, and aid management and environmental planning in line with legal and government policy obligations.
Highlights A system for remote surveying of UK roadside verges was developed. Street-view imagery was used to train a deep learning classifier. The model can identify roadside verges with high conservation potential. The automated system reduces the need for manual field surveys.
DeepVerge: Classification of roadside verge biodiversity and conservation potential
Perrett, Andrew (author) / Pollard, Harry (author) / Barnes, Charlie (author) / Schofield, Mark (author) / Qie, Lan (author) / Bosilj, Petra (author) / Brown, James M. (author)
2023-03-24
Article (Journal)
Electronic Resource
English
VERGE COVER, VERGE STRUCTURE, AND MANUFACTURING METHOD OF VERGE COVER
| European Patent Office | 2017
VERGE COVER, VERGE STRUCTURE, AND MANUFACTURING METHOD OF VERGE COVER
| European Patent Office | 2017