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

TL;DR

This paper demonstrates that deep learning models can be optimized for real-time leaf disease classification on edge devices using thermal imaging, enabling faster inference without sacrificing accuracy.

Contribution

It introduces a thermal image dataset and evaluates optimized deep learning models for efficient, real-time plant disease detection on resource-constrained edge hardware.

Findings

Models achieve up to 2.13x faster inference on edge devices.

Pruning and quantization reduce inference time while maintaining accuracy.

Edge computing enables real-time disease classification in agriculture.

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.