Automated detection, morphometric measurement, and size-class determination of harbor seals (Phoca vitulina) from UAV orthomosaics: a cascaded deep-learning pipeline with stratified validation across platforms, sensors, altitudes, and substrates
- 1NPWC
Abstract
Reliable demographic data are critical for seal population assessment and management, yet remain challenging to obtain from drone surveys without automated tools. Using data from harbor seal (Phoca vitulina) haulout surveys conducted in British Columbia, Canada, in 2023-2024, we developed and tested an automated pipeline that integrates detection, morphometric extraction, and size-class classification of seals into a single system. The pipeline combines cascaded DeepLabV3+ segmentation (water removal, colony-level seal detection, and individual body-contour refinement), principal component analysis (PCA)-based extraction of body length and maximum width, and three-class size-class determination based on body length (pups, juveniles, subadults/adults). We evaluated the pipeline on 2,483 manually annotated seals across four drone platforms (DJI Mavic 3 Thermal, Mavic 2 Pro, Mini 3 Pro and Mini 2), five substrate types, and two sensor modalities (RGB, thermal infrared). Aggregate detection F1 was 0.914. Orthomosaic geometric accuracy, quantified as the RMSE of calibration-rod length on the orthomosaic, ranged from 1.0 cm to 18.0 cm depending on the platform-altitude-substrate combination, a spread that aggregate metrics would mask. Relative to manual annotation on the same orthomosaics, automated RGB body lengths showed near-zero bias (-0.6 cm; RMSE = 12.5 cm), whereas automated thermal lengths were systematically shorter (bias -19.9 cm, RMSE = 24.1 cm), establishing the unsuitability of thermal imagery for automated quantitative pinniped morphometry. For RGB imagery, threshold classification by body length reached an overall accuracy (OA) of 78.7%; for individuals outside the uncertainty zones (defined as {+/-}2 x the geometric RMSE on calibration rods around each class boundary), OA reached 90.9%. Within the RGB subset, OA was 78.9% at 20 m altitude, declining to 75.0% at 40 m where geometric error dominated. The pipeline enables automated demographic monitoring of pinniped haulouts with quantitatively characterized uncertainty across survey conditions.
Related
- The ViRA-L (Visual Recognition Assessment-Leuven): Development, Validation, and Age Stratified Normative Characterization of a Multi-Category Visual Recognition Battery
- PPNet: a deep learning framework for rice seedling localization and distribution from UAV orthomosaics
- Individualized morphometric-similarity deviations in autism linked to cortical hierarchy and microarchitecture
- The invasion of the red dwarf honey bee (Apis florea) in Egypt: morphometric assessment and nesting preferences in urban environments
- Variation in Hygienic Behavior and Its Associations with Colony-Performance and Wing Morphometric Traits Among Honey Bee Colonies from Five Apiaries in Kazakhstan
BibTeX
@misc{usatov2026automated,
title = {{Automated detection, morphometric measurement, and size-class determination of harbor seals (Phoca vitulina) from UAV orthomosaics: a cascaded deep-learning pipeline with stratified validation across platforms, sensors, altitudes, and substrates}},
author = {I. Usatov and V. Burkanov and I. Trukhanova and A. Altukhov and A. Bowker and C. Nordstrom and S. Tucker},
howpublished = {bioRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.01.755886},
url = {https://www.biorxiv.org/content/10.64898/2026.10.01.755886v1},
note = {CC0},
}
RIS
TY - UNPB
TI - Automated detection, morphometric measurement, and size-class determination of harbor seals (Phoca vitulina) from UAV orthomosaics: a cascaded deep-learning pipeline with stratified validation across platforms, sensors, altitudes, and substrates
AU - I. Usatov
AU - V. Burkanov
AU - I. Trukhanova
AU - A. Altukhov
AU - A. Bowker
AU - C. Nordstrom
AU - S. Tucker
PY - 2026
DA - 2026/10/01
PB - bioRxiv
DO - 10.64898/2026.10.01.755886
UR - https://www.biorxiv.org/content/10.64898/2026.10.01.755886v1
AB - Reliable demographic data are critical for seal population assessment and management, yet remain challenging to obtain from drone surveys without automated tools. Using data from harbor seal (Phoca vitulina) haulout surveys conducted in British Columbia, Canada, in 2023-2024, we developed and tested an automated pipeline that integrates detection, morphometric extraction, and size-class classification of seals into a single system. The pipeline combines cascaded DeepLabV3+ segmentation (water removal, colony-level seal detection, and individual body-contour refinement), principal component analysis (PCA)-based extraction of body length and maximum width, and three-class size-class determination based on body length (pups, juveniles, subadults/adults). We evaluated the pipeline on 2,483 manually annotated seals across four drone platforms (DJI Mavic 3 Thermal, Mavic 2 Pro, Mini 3 Pro and Mini 2), five substrate types, and two sensor modalities (RGB, thermal infrared). Aggregate detection F1 was 0.914. Orthomosaic geometric accuracy, quantified as the RMSE of calibration-rod length on the orthomosaic, ranged from 1.0 cm to 18.0 cm depending on the platform-altitude-substrate combination, a spread that aggregate metrics would mask. Relative to manual annotation on the same orthomosaics, automated RGB body lengths showed near-zero bias (-0.6 cm; RMSE = 12.5 cm), whereas automated thermal lengths were systematically shorter (bias -19.9 cm, RMSE = 24.1 cm), establishing the unsuitability of thermal imagery for automated quantitative pinniped morphometry. For RGB imagery, threshold classification by body length reached an overall accuracy (OA) of 78.7%; for individuals outside the uncertainty zones (defined as {+/-}2 x the geometric RMSE on calibration rods around each class boundary), OA reached 90.9%. Within the RGB subset, OA was 78.9% at 20 m altitude, declining to 75.0% at 40 m where geometric error dominated. The pipeline enables automated demographic monitoring of pinniped haulouts with quantitatively characterized uncertainty across survey conditions.
ER -