Edge-Optimized Deep Learning Architectures for Classification of Agricultural Insects with Mobile Deployment

Information, Год журнала: 2025, Номер 16(5), С. 348 - 348

Опубликована: Апрель 25, 2025

The deployment of machine learning models on mobile platforms has ushered in a new era innovation across diverse sectors, including agriculture, where such applications hold immense promise for empowering farmers with cutting-edge technologies. In this context, the threat posed by insects to crop yields during harvest escalated, fueled factors as evolution and climate change-induced shifts insect behavior. To address challenge, smart monitoring systems detection have emerged crucial tools IoT-based systems, enabling interventions safeguard crops. primary contribution study lies its systematic investigation model optimization techniques edge deployment, Post-Training Quantization, Quantization-Aware Training, Data Representative Quantization. As such, we need efficient, on-site pest agricultural settings. We provide detailed analysis trade-offs between size, inference speed, accuracy different approaches, ensuring practical applicability resource-constrained farming environments. Our explores various methodologies development, utilization Mobile-ViT EfficientNet architectures, coupled transfer fine-tuning techniques. Using Dangerous Farm Insects Dataset, achieve an 82.6% 77.8% validation test datasets, respectively, showcasing efficacy our approach. Furthermore, investigate quantization optimize performance on-device inference, seamless devices other without compromising accuracy. best quantized model, produced through was able maintain classification while significantly reducing size from 33 MB 9.6 MB. validate generalizability solution, extended experiments larger IP102 dataset. using Quantization 59.6% also MB, thus demonstrating that solution maintains competitive broader range classes.

Язык: Английский

Unveiling host-seeking behaviour in entomopathogenic nematodes via lab-on-a-chip technology

Biosystems Engineering, Год журнала: 2025, Номер 255, С. 104159 - 104159

Опубликована: Май 9, 2025

Язык: Английский

Automated detection of larval stages of the Black Soldier Fly (Hermetia illucens Linnaeus) through deep learning augmented with optical flow

Information Processing in Agriculture, Год журнала: 2025, Номер unknown

Опубликована: Май 1, 2025

Язык: Английский