Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Low-Cost UAV Deep Learning Pipeline for Integrated Apple Disease Diagnosis,Freshness Assessment, and Fruit Detection

Soham Dutta, Soham Banerjee, Sneha Mahata, Anindya Sen, Sayantani Datta

TL;DR

The paper tackles the lack of integrated, cost-effective orchard monitoring by proposing a unified RGB-only UAV pipeline that performs leaf-disease detection, apple freshness classification, and apple localization entirely on-device using ESP32-CAM and Raspberry Pi. It deploys ResNet50, VGG16, and YOLOv8 for the three tasks, achieving high accuracies—approximately 98.9% for disease, 97.4% for freshness, and a 0.857 F1-score for detection—while avoiding cloud reliance. This approach demonstrates that a carefully staged, edge-optimized architecture can deliver real-time, multi-task orchard intelligence on affordable hardware, offering a scalable alternative to multispectral UAV solutions. The work has practical implications for resource-limited growers by enabling offline, on-site inference and easy deployment across diverse orchard settings.

Abstract

Apple orchards require timely disease detection, fruit quality assessment, and yield estimation, yet existing UAV-based systems address such tasks in isolation and often rely on costly multispectral sensors. This paper presents a unified, low-cost RGB-only UAV-based orchard intelligent pipeline integrating ResNet50 for leaf disease detection, VGG 16 for apple freshness determination, and YOLOv8 for real-time apple detection and localization. The system runs on an ESP32-CAM and Raspberry Pi, providing fully offline on-site inference without cloud support. Experiments demonstrate 98.9% accuracy for leaf disease classification, 97.4% accuracy for freshness classification, and 0.857 F1 score for apple detection. The framework provides an accessible and scalable alternative to multispectral UAV solutions, supporting practical precision agriculture on affordable hardware.

A Low-Cost UAV Deep Learning Pipeline for Integrated Apple Disease Diagnosis,Freshness Assessment, and Fruit Detection

TL;DR

The paper tackles the lack of integrated, cost-effective orchard monitoring by proposing a unified RGB-only UAV pipeline that performs leaf-disease detection, apple freshness classification, and apple localization entirely on-device using ESP32-CAM and Raspberry Pi. It deploys ResNet50, VGG16, and YOLOv8 for the three tasks, achieving high accuracies—approximately 98.9% for disease, 97.4% for freshness, and a 0.857 F1-score for detection—while avoiding cloud reliance. This approach demonstrates that a carefully staged, edge-optimized architecture can deliver real-time, multi-task orchard intelligence on affordable hardware, offering a scalable alternative to multispectral UAV solutions. The work has practical implications for resource-limited growers by enabling offline, on-site inference and easy deployment across diverse orchard settings.

Abstract

Apple orchards require timely disease detection, fruit quality assessment, and yield estimation, yet existing UAV-based systems address such tasks in isolation and often rely on costly multispectral sensors. This paper presents a unified, low-cost RGB-only UAV-based orchard intelligent pipeline integrating ResNet50 for leaf disease detection, VGG 16 for apple freshness determination, and YOLOv8 for real-time apple detection and localization. The system runs on an ESP32-CAM and Raspberry Pi, providing fully offline on-site inference without cloud support. Experiments demonstrate 98.9% accuracy for leaf disease classification, 97.4% accuracy for freshness classification, and 0.857 F1 score for apple detection. The framework provides an accessible and scalable alternative to multispectral UAV solutions, supporting practical precision agriculture on affordable hardware.
Paper Structure (11 sections, 4 figures, 1 table)

This paper contains 11 sections, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Literature Review
  3. 1.UAV-based Image acquisition:
  4. 2.Edge Processing:
  5. 3.Three vision tasks:
  6. Methodology
  7. Results
  8. Discussion:
  9. Comparative analysis with existing literature
  10. Conclusion
  11. Future scope

Figures (4)

  • Figure 1: Proposed pipeline
  • Figure 2: Leaf-disease predictions from ResNet-50, correctly classifying healthy, black rot, cedar apple rust, and apple scab samples.
  • Figure 3: Freshness classification outputs from VGG16, correctly identifying a rotten apple (left) and a fresh apple (right) with high confidence
  • Figure 4: Apple detection results from YOLOv8, accurately localizing apples under varying lighting and occlusion conditions.