Automatic detection, tracking and counting of birds in marine video content
T'Jampens, R.; Hernandez, F.; Vandecasteele, F.; Verstockt, S. (2016). Automatic detection, tracking and counting of birds in marine video content, in: López, M.B. et al. IPTA 2016: 6th International Conference on Image Processing Theory, Tools and Applications, Oulu, Finland, December 2016. International Conference on Image Processing Theory Tools and Applications, 6: pp. [1-6]. https://dx.doi.org/10.1109/IPTA.2016.7821031 In: López, M.B. et al. (Ed.) (2016). IPTA 2016: 6th International Conference on Image Processing Theory, Tools and Applications, Oulu, Finland, December 2016. International Conference on Image Processing Theory Tools and Applications, 6. IEEE: New York. ISBN 978-1-4673-8910-5. , more In: International Conference on Image Processing Theory Tools and Applications. IEEE: New York. ISSN 2154-512X, more |  |
| Available in | Authors | | Document type: Conference paper
|
| Keywords | Aquatic organisms > Marine organisms > Aquatic birds > Marine birds
Detection
Aves [WoRMS]
Marine/Coastal | | Author keywords | dynamic background subtraction; texture analysis; image classification; object detection; seabirds |
| Abstract | Robust automatic detection of moving objects in a marine context is a multi-faceted problem due to the complexity of the observed scene. The dynamic nature of the sea caused by waves, boat wakes, and weather conditions poses huge challenges for the development of a stable background model. Moreover, camera motion, reflections, lightning and illumination changes may contribute to false detections. Dynamic background subtraction (DBGS) is widely considered as a solution to tackle this issue in the scope of vessel detection for maritime traffic analysis. In this paper, the DBGS techniques suggested for ships are investigated and optimized for the monitoring and tracking of birds in marine video content. In addition to background subtraction, foreground candidates are filtered by a classifier based on their feature descriptors in order to remove non-bird objects. Different types of classifiers have been evaluated and results on a ground truth labeled dataset of challenging video fragments show similar levels of precision and recall of about 95% for the best performing classifier. The remaining foreground items are counted and birds are tracked along the video sequence using spatio-temporal motion prediction. This allows marine scientists to study the presence and behavior of birds. |
|
