Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys
Abstract Imagery from drones is becoming common in wildlife research and management, but processing data efficiently remains a challenge. We developed a methodology for training a convolutional neural network model on large-scale mosaic imagery to detect and count caribou (Rangifer tarandus), compar...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2023-01-01
|
| Series: | Scientific Reports |
| Online Access: | https://doi.org/10.1038/s41598-023-28240-9 |
| _version_ | 1797945987791060992 |
|---|---|
| author | Javier Lenzi Andrew F. Barnas Abdelrahman A. ElSaid Travis Desell Robert F. Rockwell Susan N. Ellis-Felege |
| author_facet | Javier Lenzi Andrew F. Barnas Abdelrahman A. ElSaid Travis Desell Robert F. Rockwell Susan N. Ellis-Felege |
| author_sort | Javier Lenzi |
| collection | DOAJ |
| description | Abstract Imagery from drones is becoming common in wildlife research and management, but processing data efficiently remains a challenge. We developed a methodology for training a convolutional neural network model on large-scale mosaic imagery to detect and count caribou (Rangifer tarandus), compare model performance with an experienced observer and a group of naïve observers, and discuss the use of aerial imagery and automated methods for large mammal surveys. Combining images taken at 75 m and 120 m above ground level, a faster region-based convolutional neural network (Faster-RCNN) model was trained in using annotated imagery with the labels: “adult caribou”, “calf caribou”, and “ghost caribou” (animals moving between images, producing blurring individuals during the photogrammetry processing). Accuracy, precision, and recall of the model were 80%, 90%, and 88%, respectively. Detections between the model and experienced observer were highly correlated (Pearson: 0.96–0.99, P value < 0.05). The model was generally more effective in detecting adults, calves, and ghosts than naïve observers at both altitudes. We also discuss the need to improve consistency of observers’ annotations if manual review will be used to train models accurately. Generalization of automated methods for large mammal detections will be necessary for large-scale studies with diverse platforms, airspace restrictions, and sensor capabilities. |
| first_indexed | 2024-04-10T21:03:45Z |
| format | Article |
| id | doaj.art-ed21980fda49496a81a363156555dc22 |
| institution | Directory Open Access Journal |
| issn | 2045-2322 |
| language | English |
| last_indexed | 2024-04-10T21:03:45Z |
| publishDate | 2023-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj.art-ed21980fda49496a81a363156555dc222023-01-22T12:12:43ZengNature PortfolioScientific Reports2045-23222023-01-0113111310.1038/s41598-023-28240-9Artificial intelligence for automated detection of large mammals creates path to upscale drone surveysJavier Lenzi0Andrew F. Barnas1Abdelrahman A. ElSaid2Travis Desell3Robert F. Rockwell4Susan N. Ellis-Felege5Department of Biology, University of North DakotaDepartment of Biology, University of North DakotaDepartment of Computer Science, University of North Carolina WilmingtonDepartment of Software Engineering, Rochester Institute of TechnologyVertebrate Zoology, American Museum of Natural HistoryDepartment of Biology, University of North DakotaAbstract Imagery from drones is becoming common in wildlife research and management, but processing data efficiently remains a challenge. We developed a methodology for training a convolutional neural network model on large-scale mosaic imagery to detect and count caribou (Rangifer tarandus), compare model performance with an experienced observer and a group of naïve observers, and discuss the use of aerial imagery and automated methods for large mammal surveys. Combining images taken at 75 m and 120 m above ground level, a faster region-based convolutional neural network (Faster-RCNN) model was trained in using annotated imagery with the labels: “adult caribou”, “calf caribou”, and “ghost caribou” (animals moving between images, producing blurring individuals during the photogrammetry processing). Accuracy, precision, and recall of the model were 80%, 90%, and 88%, respectively. Detections between the model and experienced observer were highly correlated (Pearson: 0.96–0.99, P value < 0.05). The model was generally more effective in detecting adults, calves, and ghosts than naïve observers at both altitudes. We also discuss the need to improve consistency of observers’ annotations if manual review will be used to train models accurately. Generalization of automated methods for large mammal detections will be necessary for large-scale studies with diverse platforms, airspace restrictions, and sensor capabilities.https://doi.org/10.1038/s41598-023-28240-9 |
| spellingShingle | Javier Lenzi Andrew F. Barnas Abdelrahman A. ElSaid Travis Desell Robert F. Rockwell Susan N. Ellis-Felege Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys Scientific Reports |
| title | Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys |
| title_full | Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys |
| title_fullStr | Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys |
| title_full_unstemmed | Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys |
| title_short | Artificial intelligence for automated detection of large mammals creates path to upscale drone surveys |
| title_sort | artificial intelligence for automated detection of large mammals creates path to upscale drone surveys |
| url | https://doi.org/10.1038/s41598-023-28240-9 |
| work_keys_str_mv | AT javierlenzi artificialintelligenceforautomateddetectionoflargemammalscreatespathtoupscaledronesurveys AT andrewfbarnas artificialintelligenceforautomateddetectionoflargemammalscreatespathtoupscaledronesurveys AT abdelrahmanaelsaid artificialintelligenceforautomateddetectionoflargemammalscreatespathtoupscaledronesurveys AT travisdesell artificialintelligenceforautomateddetectionoflargemammalscreatespathtoupscaledronesurveys AT robertfrockwell artificialintelligenceforautomateddetectionoflargemammalscreatespathtoupscaledronesurveys AT susannellisfelege artificialintelligenceforautomateddetectionoflargemammalscreatespathtoupscaledronesurveys |