Abstract:Deep learning (DL) technologies can transform agriculture by improving crop health monitoring and management, thus improving food safety. In this paper, we explore the potential of edge computing for real-time classification of leaf diseases using thermal imaging. We present a thermal image dataset for plant disease classification and evaluate deep learning models, including InceptionV3, MobileNetV1, MobileNetV2, and VGG-16, on resource-constrained devices like the Raspberry Pi 4B. Using pruning and quantization-aware training, these models achieve inference times up to 1.48x faster on Edge TPU Max for VGG16, and up to 2.13x faster with precision reduction on Intel NCS2 for MobileNetV1, compared to high-end GPUs like the RTX 3090, while maintaining state-of-the-art accuracy.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **Use edge computing and thermal imaging technology to achieve real - time classification of plant leaf diseases, in order to improve the efficiency of agricultural crop health monitoring and management, and thus ensure food safety**. Specifically, this research mainly focuses on the following aspects: 1. **Introduce a new dataset**: - The author introduced a new thermal imaging dataset of 7 types of leaf diseases of 16 plants, with a total of 15,444 images. These images cover a variety of common plant leaf diseases in the Sao Paulo region of Brazil. 2. **Evaluate model optimization techniques**: - Researchers evaluated the optimization effects of pruning and quantization - aware training (QAT) on deep - learning models such as MobileNetV1, MobileNetV2, VGG16 and InceptionV3, enabling them to run efficiently on resource - constrained devices (such as Raspberry Pi 4B). 3. **Develop a real - time classification solution based on hardware acceleration**: - By using hardware acceleration devices such as Edge TPU (Coral USB) and Intel NCS2, researchers developed a solution that can perform real - time image classification on edge devices. This enables rapid and accurate identification of leaf diseases even in the absence of powerful computing resources. ### Key issues and challenges - **Real - time requirements**: Traditional cloud - infrastructure methods have latency problems when dealing with real - time image classification, especially in environments with limited bandwidth or unstable network connections. - **Resource - constrained devices**: Edge devices such as Raspberry Pi 4B have limited computing resources, and deep - learning models need to be compressed and optimized to ensure their efficient operation on these devices. - **Importance of early detection**: Timely discovery and diagnosis of plant leaf diseases are crucial for reducing crop losses, especially in the context of climate change and frequent extreme weather events. ### Solutions - **Thermal imaging technology**: Compared with visible - light images, thermal imaging can overcome the limitations of lighting conditions and provide more stable and reliable disease detection results. - **Edge computing**: Transferring computing tasks from the cloud to edge devices reduces the latency and bandwidth requirements of data transmission and achieves real - time classification. - **Model compression and optimization**: Through pruning and quantization - aware training, the size and computational complexity of the model are significantly reduced while maintaining high classification accuracy. In summary, this paper aims to develop an efficient real - time leaf - disease classification system by combining thermal imaging technology and edge computing to address the challenges of leaf - disease detection in agriculture and improve crop - health management and food - safety levels.

An Edge Computing-Based Solution for Real-Time Leaf Disease Classification using Thermal Imaging

tCrop: Thermal Imaging Based Plant Stress Identification Using On-Edge Deep Learning

Deep Learning-Based Leaf Disease Detection in Crops Using Images for Agricultural Applications

Plant Disease Identification Using Machine Learning Algorithms on Single-Board Computers in IoT Environments

Enhancing agriculture through real-time grape leaf disease classification via an edge device with a lightweight CNN architecture and Grad-CAM

Using Mobile Edge AI to Detect and Map Diseases in Citrus Orchards

Real-time plant health assessment via implementing cloud-based scalable transfer learning on AWS DeepLens

A Novel Feature Extraction Model for the Detection of Plant Disease from Leaf Images in Low Computational Devices

AgroLens: A low-cost and green-friendly Smart Farm Architecture to support real-time leaf disease diagnostics

Deep Learning based Leaf Disease Detection using Convolutional Neural Network

Deep Learning-Based Leaf Disease Detection in Crop Using Images for Agricultural Application

Less is More: Lighter and Faster Deep Neural Architecture for Tomato Leaf Disease Classification

ELECTRICAL EXPRESSION OF SPIN ACCUMULATION IN FERROMAGNET/SEMICONDUCTOR STRUCTURES

Deep Convolutional Neural Network based Detection System for Real-time Corn Plant Disease Recognition

Smart Disease Detection System for Citrus Fruits Using Deep Learning with Edge Computing

Optimized deep learning network for plant leaf disease segmentation and multi-classification using leaf images

Streamlining plant disease diagnosis with convolutional neural networks and edge devices

Lightweight Inception Networks for the Recognition and Detection of Rice Plant Diseases

Plant Disease Diagnosis and Image Classification Using Deep Learning

Performance evaluation of edge-computing platforms for the prediction of low temperatures in agriculture using deep learning

Deep learning-based automated disease detection and classification model for precision agriculture