TinyEcoWeedNet: Edge Efficient Real-Time Aerial Agricultural Weed Detection

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Khater, Omar H., Siddiqui, Abdul Jabbar, El-Maleh, Aiman, Hossain, M. Shamim
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915506943950848
author Khater, Omar H.
Siddiqui, Abdul Jabbar
El-Maleh, Aiman
Hossain, M. Shamim
author_facet Khater, Omar H.
Siddiqui, Abdul Jabbar
El-Maleh, Aiman
Hossain, M. Shamim
contents Deploying deep learning models in agriculture is difficult because edge devices have limited resources, but this work presents a compressed version of EcoWeedNet using structured channel pruning, quantization-aware training (QAT), and acceleration with NVIDIA's TensorRT on the Jetson Orin Nano. Despite the challenges of pruning complex architectures with residual shortcuts, attention mechanisms, concatenations, and CSP blocks, the model size was reduced by up to 68.5% and computations by 3.2 GFLOPs, while inference speed reached 184 FPS at FP16, 28.7% faster than the baseline. On the CottonWeedDet12 dataset, the pruned EcoWeedNet with a 39.5% pruning ratio outperformed YOLO11n and YOLO12n (with only 20% pruning), achieving 83.7% precision, 77.5% recall, and 85.9% mAP50, proving it to be both efficient and effective for precision agriculture.
format Preprint
id arxiv_https___arxiv_org_abs_2509_18193
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle TinyEcoWeedNet: Edge Efficient Real-Time Aerial Agricultural Weed Detection
Khater, Omar H.
Siddiqui, Abdul Jabbar
El-Maleh, Aiman
Hossain, M. Shamim
Computer Vision and Pattern Recognition
Artificial Intelligence
Deploying deep learning models in agriculture is difficult because edge devices have limited resources, but this work presents a compressed version of EcoWeedNet using structured channel pruning, quantization-aware training (QAT), and acceleration with NVIDIA's TensorRT on the Jetson Orin Nano. Despite the challenges of pruning complex architectures with residual shortcuts, attention mechanisms, concatenations, and CSP blocks, the model size was reduced by up to 68.5% and computations by 3.2 GFLOPs, while inference speed reached 184 FPS at FP16, 28.7% faster than the baseline. On the CottonWeedDet12 dataset, the pruned EcoWeedNet with a 39.5% pruning ratio outperformed YOLO11n and YOLO12n (with only 20% pruning), achieving 83.7% precision, 77.5% recall, and 85.9% mAP50, proving it to be both efficient and effective for precision agriculture.
title TinyEcoWeedNet: Edge Efficient Real-Time Aerial Agricultural Weed Detection
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2509.18193