1PG Scholar, Department of Electronics and Telecommunication Engineering, Ashokrao Mane Group of Institutions, Vathar, Kolhapur, Maharashtra, India.
2Associate Professor, Department of Electronics and Telecommunication Engineering, Ashokrao Mane Group of Institutions, Vathar, Kolhapur, Maharashtra, India.
*Corresponding Author: Rohit R. Vibhute, Email: rohitvibhute040@gmail.com; Dr. Seema S. Patil, Email: seema.patil9@gmail.com
The structure of the plant canopy seems to play a very important role in shaping the microclimate around the crops. This is done by influencing the evaporation, light interception, and overall crop growth, which greatly depends on the canopy. It is also a very important indicator of biomass accumulation and directly related to yield potation. Due to this accurate information about the canopy presence and the size of the canopy is very much essential for environmental planning. The conventional approaches which are implemented for canopy detection such as manual field surveys and hemispherical photography are usually time-consuming and labour-intensive. This means they are often affected by the observer bias. Although various techniques such as remote sensing, which use satellite imagery or LiDAR exist and can cover large areas, they are typically limited due to spatial resolution and high operational costs. Over the recent years, we have seen a large technological progress in the field of deep learning and computer vision. This has opened infinite new possibilities for automated canopy analysis using ground-level images. In this paper, we discuss a lightweight and efficient framework for plant canopy detection and size estimation. A mobile net SSD-based convolutional neural network is fine-tuned using diverse plant images dataset to accurately detect canopy regions. The canopy size is then estimated for detection based on the pixel area of the detected regions. To support the practical deployment, the system consists of an interactive graphical interface. The experiment evaluation which is carried out shows reliable detection accuracy with low computational requirement, thus making the proposed approach suitable for real-time on-field applications. The important key observations which can be observed from this study include a comprehensive review of related work, end-to-end methodology, and covering model training with results and discussion along with the future directions.
Keywords: Tree Canopy Detection, Deep Learning, Convolutional Neural Network (CNN), Artificial Intelligence, MobileNet-SSD, Canopy Size Estimation
How to cite this article: Vibhute RR, Patil SS. Tree Canopy Detection Using Artificial Intelligence Based - Deep Learning Approach. Int J Drug Deliv Technol. 2026;16(11s): 223-231; DOI: 10.25258/ijddt.16.11s.21
Source of support: Nil.
Conflict of interest: Nil