Balancing complexity and performance of machine learning models for avian pests sound detection in agricultural environments

dc.contributor.authorKazeneza, Micheline
dc.contributor.authorBosman, Anna Sergeevna
dc.contributor.authorAmenyedzi, Destiny Kwabla
dc.contributor.authorHanyurwimfura, Damien
dc.contributor.authorNdashimye, Emmanuel
dc.contributor.authorVodacek, Anthony
dc.date.accessioned2025-07-29T11:26:37Z
dc.date.available2025-07-29T11:26:37Z
dc.date.issued2025-06
dc.description.abstractAgricultural pest control traditionally relies on inefficient visual inspections. Acoustic monitoring offers a promising alternative by analyzing pest-specific sounds. While effective, implementing acoustic monitoring in agricultural settings faces practical constraints, particularly the limited computational resources available in remote farming environments. This necessitates optimized machine learning (ML) solutions for low-power edge devices. This study evaluates ML models for bird pest detection on resource-constrained platforms. We evaluated convolutional neural networks (CNNs), recurrent neural networks (RNNs), and traditional ML models by comparing standalone and knowledge-distilled versions of EfficientNetB0 and gated recurrent unity (GRU) against EfficientNetB4, Long short-term memory (LSTM), MobileNetV2, LightGBM, and support vector machine (SVM). Analysis revealed significant performance variations across computational requirements. LightGBM achieved 98% accuracy with minimal resources (8,500 parameters, 7KB, 0.6ms inference), demonstrating good efficiency. SVM (97% accuracy) and distilled GRU (86% accuracy) also showed favorable performance-to-resource ratios. Knowledge distillation substantially enhanced the accuracy of EfficientNetB0 (from 73% to 98%) and modestly improved GRU (from 84% to 86%). We examined platform compatibility across computing tiers, discovering that while high-performance edge devices (Jetson Nano, Raspberry Pi 4) support all studied models effectively, microcontrollers require specialized approaches. Advanced microcontrollers (such as ESP32-S3 and STM32H7) can accommodate optimized implementations, while highly constrained platforms (such as Arduino Nano) require TinyML techniques. This research contributes 1) an on-farm audio dataset, 2) comprehensive cross-model evaluation metrics, and 3) deployment optimization strategies for acoustic pest detection systems in resource-constrained agricultural environments.
dc.description.departmentComputer Science
dc.description.librarianhj2025
dc.description.sdgSDG-02: Zero Hunger
dc.description.sdgSDG-09: Industry, innovation and infrastructure
dc.description.sponsorshipThis work was supported in part by the icipe-World Bank Financing, in part by the World Bank-Korea Trust Fund for Partnership for Skills in Applied Sciences, Engineering and Technology (PASET)-Regional Scholarship and Innovation Fund (RSIF), and in part by the University of Pretoria through PASET-RSIF Research Funds.
dc.description.urihttp://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=6287639
dc.identifier.citationM. Kazeneza, A.S. Bosman, D.K. Amenyedzi, D. Hanyurwimfura, E. Ndashimye and A. Vodacek, "Balancing Complexity and Performance of Machine Learning Models for Avian Pests Sound Detection in Agricultural Environments," in IEEE Access, vol. 13, pp. 105813-105827, 2025, doi: 10.1109/ACCESS.2025.3580620.
dc.identifier.issn2169-3536 (online)
dc.identifier.other10.1109/ACCESS.2025.3580620
dc.identifier.urihttp://hdl.handle.net/2263/103663
dc.language.isoen
dc.publisherInstitute of Electrical and Electronics Engineers Inc.
dc.rights© 2025 The Authors. This work is licensed under a Creative Commons Attribution 4.0 License. See https://creativecommons.org/licenses/by/4.0.
dc.subjectMachine learning
dc.subjectKnowledge distillation
dc.subjectDeep learning
dc.subjectBird pest detection
dc.subjectAcoustics
dc.subjectComputational modeling
dc.subjectRecording
dc.subjectComplexity theory
dc.subjectAccuracy
dc.subjectMonitoring
dc.subjectSpectrogram
dc.subjectComputer architecture
dc.subjectAcoustic monitoring
dc.subjectAgricultural pest control
dc.titleBalancing complexity and performance of machine learning models for avian pests sound detection in agricultural environments
dc.typeArticle

Files

Original bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Kazeneza_Balancing_2025.pdf
Size:
6.13 MB
Format:
Adobe Portable Document Format
Description:
Article

License bundle

Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description: