Improving Bird Classification with Primary Color Additives

2025-07-24

AI Summary

To address degraded avian audio classification performance under low signal-to-noise ratios, overlapping multi-species vocalizations, and label scarcity, this paper proposes a Chroma-Enhanced Spectrogram method. It encodes biologically salient acoustic features—including pitch, rhythm, and repetitive motifs—into chromatic channels, augmenting conventional spectrograms with an additional color-coded dimension to emphasize species-specific spectral patterns. By integrating audio signal processing with visual representation learning, the method enhances deep models’ robustness in discriminating overlapping bird vocalization motifs. Evaluated on the BirdCLEF 2024 benchmark, it achieves absolute improvements of 7.3% in F1-score, 6.2% in ROC-AUC, and 6.6% in cMAP over the baseline grayscale spectrogram baseline, outperforming the competition’s winning solution. These results empirically validate that chromatic encoding significantly improves the discriminability of acoustic features in challenging real-world avian soundscapes.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Computer Vision: Multi-modal Vision

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

We address the problem of classifying bird species using their song recordings, a challenging task due to environmental noise, overlapping vocalizations, and missing labels. Existing models struggle with low-SNR or multi-species recordings. We hypothesize that birds can be classified by visualizing their pitch pattern, speed, and repetition, collectively called motifs. Deep learning models applied to spectrogram images help, but similar motifs across species cause confusion. To mitigate this, we embed frequency information into spectrograms using primary color additives. This enhances species distinction and improves classification accuracy. Our experiments show that the proposed approach achieves statistically significant gains over models without colorization and surpasses the BirdCLEF 2024 winner, improving F1 by 7.3%, ROC-AUC by 6.2%, and CMAP by 6.6%. These results demonstrate the effectiveness of incorporating frequency information via colorization.

Problem

• Classifying bird species using noisy song recordings
• Distinguishing similar motifs across different bird species
• Improving classification accuracy with color-enhanced spectrograms

Innovation

• Use primary color additives in spectrograms
• Embed frequency information for species distinction
• Improve classification accuracy with colorization