PalmsCNN: Palm Tree Detection

Version: 1.3.0 | Institution: Instituto de Investigaciones de la Amazonía Peruana (IIAP)

Introduction

PalmsCNN is a QGIS plugin designed to automatically detect three Amazonian palm species —aguaje (Mauritia flexuosa), huasaí (Euterpe precatoria), and ungurahui (Oenocarpus bataua)— in high-resolution RGB drone imagery. The system uses a convolutional neural network (CNN) model to identify and map palm trees across Amazonian landscapes.

Installation and Dependencies

1. When you run the plugin for the first time, it will automatically create a virtual environment (venv) containing all required libraries: numpy, scikit-image, and onnxruntime.
2. This process requires Internet access only during the first setup.
3. Once the environment is ready, it will be stored locally and reused in future sessions.

Usage

1. In QGIS, open Processing → PalmsCNN: Palm Tree Detection.
2. Select an RGB image (drone or high-resolution satellite imagery).
3. Choose the target species or select “All”.
4. Run the algorithm and wait for the detection results.
5. The output layer will display the georeferenced locations of detected palms.

Troubleshooting

• Virtual environment not created: Check Internet connection and write permissions in the user folder.
• SSL error: Ensure your QGIS Python interpreter has SSL support (default on Windows and Linux builds).
• Plugin not visible: Make sure you are using QGIS version 3.34 or later.

Credits

Developed by the Artificial Intelligence Laboratory - BOSQUES – IIAP, Peru.

Authors: Susan Palacios, Rodolfo Cardenas, Stephano Torres, Ximena Tagle

Contact: rcardenasv@iiap.gob.pe