Towards Automated Biometric Identification of Sea Turtles (Chelonia mydas)

Abstract

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

Chelonia mydas detection and image extraction from noisy field recordings

“…The process becomes more time consuming as more and more images are extracted from the videos or photos [13]. Other than that, some images of wildlife captured using motion-based sensors contains background elements that are insignificant to the research and are beyond the region of interest (ROI) needed [14]. The reliability of the trap cameras to operationally collect only the subject animal may also be reduced because of the trap cameras local environment such as the presence of snowfall or wind which can trigger the trap cameras automatic sensors, thus generating false positives [15].…”

Performance evaluation of SIFT against common image deformations on iban plaited mat motif images

“…To date, a significant amount of research had been done to perform robust detection, matching and recognition of discriminative features points inside an image. Feature detectors and matching algorithms had been developed for various purposes, [8]- [12], features recognition [13]- [16], and gesture recognition [17], among others. This work aims to evaluate the performance of SIFT [18] against common image deformations applied to a clean synthetic mat motif image.…”