Balancing Domain Experts for Long-Tailed Camera-Trap Recognition

TL;DR

This paper introduces a unified framework for long-tailed camera-trap recognition, addressing data imbalance with domain experts and flow consistency loss, validated on two new datasets with improved performance.

Contribution

The paper proposes a novel framework with domain experts and flow consistency loss tailored for long-tailed camera-trap data, along with two new datasets for evaluation.

Findings

Outperforms previous methods on recessive domain samples

Effectively balances decision boundaries across domains

Enhances focus on moving objects using flow consistency loss

Abstract

Label distributions in camera-trap images are highly imbalanced and long-tailed, resulting in neural networks tending to be biased towards head-classes that appear frequently. Although long-tail learning has been extremely explored to address data imbalances, few studies have been conducted to consider camera-trap characteristics, such as multi-domain and multi-frame setup. Here, we propose a unified framework and introduce two datasets for long-tailed camera-trap recognition. We first design domain experts, where each expert learns to balance imperfect decision boundaries caused by data imbalances and complement each other to generate domain-balanced decision boundaries. Also, we propose a flow consistency loss to focus on moving objects, expecting class activation maps of multi-frame matches the flow with optical flow maps for input images. Moreover, two long-tailed camera-trap datasets, WCS-LT and DMZ-LT, are introduced to validate our methods. Experimental results show the effectiveness of our framework, and proposed methods outperform previous methods on recessive domain samples.