# Detection and Identification of Penguins Using Appearance and Motion Features

arXiv: https://arxiv.org/abs/2603.03603 · HTML: https://mlsanity.com/paper/2603.03603/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 16 references that are in the map.
- First posted on arXiv 4 Mar 2026.

## Builds on (most cited first, 9 of 9)

- [Mask R-CNN](https://mlsanity.com/paper/1703.06870/) (2017)
- [YOLOv11: An Overview of the Key Architectural Enhancements](https://mlsanity.com/paper/2410.17725/) (2024)
- [Flow-Guided Feature Aggregation for Video Object Detection](https://mlsanity.com/paper/1703.10025/) (2017)
- [Learning from Failure: Training Debiased Classifier from Biased Classifier](https://mlsanity.com/paper/2007.02561/) (2020)
- [TransVOD: End-to-End Video Object Detection with Spatial-Temporal Transformers](https://mlsanity.com/paper/2201.05047/) (2022)
- [Adapting Grad-CAM for Embedding Networks](https://mlsanity.com/paper/2001.06538/) (2020)
- [YOLOv8 to YOLO11: A Comprehensive Architecture In-depth Comparative Review](https://mlsanity.com/paper/2501.13400/) (2025)
- [Lightweight Multi-Frame Integration for Robust YOLO Object Detection in Videos](https://mlsanity.com/paper/2506.20550/) (2025)
- [A Simple Detector with Frame Dynamics is a Strong Tracker](https://mlsanity.com/paper/2505.04917/) (2025)

## Abstract (arXiv)

In animal facilities, continuous surveillance of penguins is essential yet technically challenging due to their homogeneous visual characteristics, rapid and frequent posture changes, and substantial environmental noise such as water reflections. In this study, we propose a framework that enhances both detection and identification performance by integrating appearance and motion features. For detection, we adapted YOLO11 to process consecutive frames to overcome the lack of temporal consistency in single-frame detectors. This approach leverages motion cues to detect targets even when distinct visual features are obscured. Our evaluation shows that fine-tuning the model with two-frame inputs improves mAP@0.5 from 0.922 to 0.933, outperforming the baseline, and successfully recovers individuals that are indistinguishable in static images. For identification, we introduce a tracklet-based contrastive learning approach applied after tracking. Through qualitative visualization, we demonstrate that the method produces coherent feature embeddings, bringing samples from the same individual closer in the feature space, suggesting the potential for mitigating ID switching.