Finding Nemo’s Giant Cousin: Keypoint Matching for Robust Re-Identification of Giant Sunfish

The Giant Sunfish (Mola alexandrini) has unique patterns on its body, which allow for individual identification. By continuously gathering and matching images, it is possible to monitor track individuals across location time. However, images manually a tedious time-consuming task. To automate the process, we propose pipeline based finding keypoints between image pairs. We evaluate our with four different keypoint descriptors, namely ORB, SIFT, RootSIFT, SuperPoint, demonstrate that number of pair strong indicator likelihood they contain same individual. best results are obtained achieves an mAP 75.91% test dataset (TinyMola+) without training or fine-tuning any parts pipeline. Furthermore, show generalizes other domains, such as re-identification seals cows. Lastly, discuss impracticality ranking-based output real-life tasks alternative approach by viewing binary classification. can be easily modified minimal provide precision 98% recall 44% TinyMola+ dataset, basically eliminates need manual verification nearly half dataset.