Using Convolutional Neural Network to Detect and Count Individuals on Eucalyptus Plantation

Luiz Antonio Macarini, Antonio C. Sobieranski


Deep Learning constitutes a modern approach for image processing with considerable potential and promising results. As Deep Learning has been successfully applied to various application domains, it has also recently employed
in Precision Agriculture. Taking this into account, this work proposes the use of machine learning techniques, more specifically Convolutional Neural Networks (CNN), to detect and count individuals in eucalyptus plantation images, acquired from Unmanned Aerial Vehicles (UAV). The obtained results were provided by a Faster R-CNN Resnet101, with validation procedure performed against manual human annotation. Experimental results demonstrated a overall precision of 95.77% and the affordability of the approach for forestry inventories.

Texto completo:



  • Não há apontamentos.