Plantation Monitoring Using Drone Images: A Dataset and Performance Review

2025-02-12

AI Summary

Smallholder farmers in developing countries often cannot afford high-cost agricultural monitoring systems. Method: We introduce the first open-source UAV-captured RGB image dataset specifically designed for fine-grained health assessment of individual plantation saplings, with per-plant annotations across three classes: “Healthy”, “Stunted”, and “Dead”. Our approach integrates deep convolutional feature extraction—using ResNet/VGG backbones enhanced with depthwise separable convolutions—with state-of-the-art object detection frameworks (YOLOv5 and DETR) to achieve end-to-end sapling localization and health classification. Results: Experimental evaluation demonstrates that the deep convolutional design significantly improves classification robustness, boosting accuracy by 8.3%; YOLOv5 achieves an mAP@0.5 of 72.1% on single-sapling localization. This work establishes a reproducible, low-cost, and high-accuracy foundation for plantation health monitoring, offering both a benchmark dataset and an extensible technical paradigm for resource-constrained agricultural settings.

Technology Category

• Computer Vision: Object Detection & Categorization
• Machine Learning: Hardware-aware ML
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Automatic monitoring of tree plantations plays a crucial role in agriculture. Flawless monitoring of tree health helps farmers make informed decisions regarding their management by taking appropriate action. Use of drone images for automatic plantation monitoring can enhance the accuracy of the monitoring process, while still being affordable to small farmers in developing countries such as India. Small, low cost drones equipped with an RGB camera can capture high-resolution images of agricultural fields, allowing for detailed analysis of the well-being of the plantations. Existing methods of automated plantation monitoring are mostly based on satellite images, which are difficult to get for the farmers. We propose an automated system for plantation health monitoring using drone images, which are becoming easier to get for the farmers. We propose a dataset of images of trees with three categories: ``Good health", ``Stunted", and ``Dead". We annotate the dataset using CVAT annotation tool, for use in research purposes. We experiment with different well-known CNN models to observe their performance on the proposed dataset. The initial low accuracy levels show the complexity of the proposed dataset. Further, our study revealed that, depth-wise convolution operation embedded in a deep CNN model, can enhance the performance of the model on drone dataset. Further, we apply state-of-the-art object detection models to identify individual trees to better monitor them automatically.

Problem

• Automated tree plantation monitoring
• Use of drone images for agriculture
• Enhancing CNN model performance

Innovation

• Drone images for monitoring
• CNN models for analysis
• Depth-wise convolution enhancement