Towards Automated Biometric Identification of Sea Turtles (Chelonia mydas)

Irwandi Hipiny (Faculty of Computer Science and Information Technology, UNIMAS, Jalan Datuk Mohammad Musa, 94300 Kota Samarahan, Sarawak,)
Hamimah Ujir (Faculty of Computer Science and Information Technology, UNIMAS, Jalan Datuk Mohammad Musa, 94300 Kota Samarahan, Sarawak,)
Aazani Mujahid (Faculty of Resource Science and Technology, UNIMAS, Jalan Datuk Mohammad Musa, 94300 Kota Samarahan, Sarawak,)
Nurhartini Kamalia Yahya (Danau Girang Field Centre, Sabah Wildlife Department and Cardiff University, Lower Kinabatangan Wildlife Sanctuary, Sabah,)

Journal of ICT Research and Applications
Vol. 12 No. 3 (2018)

Passive biometric identification enables wildlife monitoring with minimal disturbance. Using a motion-activated camera placed at an elevated position and facing downwards, images of sea turtle carapaces were collected, each belonging to one of sixteen Chelonia mydas juveniles. Then, co-variant and robust image descriptors from these images were learned, enabling indexing and retrieval. In this paper, several classification results of sea turtle carapaces using the learned image descriptors are presented. It was found that a template-based descriptor, i.e. Histogram of Oriented Gradients (HOG) performed much better during classification than keypoint-based descriptors. For our dataset, a high-dimensional descriptor is a must because of the minimal gradient and color information in the carapace images. Using HOG, we obtained an average classification accuracy of 65%.