Classification of invasive species in Finland with deep learning
Bahrami, Ali (2021)
Bahrami, Ali
2021
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2021060113247
https://urn.fi/URN:NBN:fi:amk-2021060113247
Tiivistelmä
In this project, an application was developed in order to classify whether a plant is lupine, hogweeds or impatiens, based on the image of the plant. These plants are considered as invasive species in Finland. The first step in order to remove an invasive plant from the environment is to recognize it, either by a human or a machine. The aim of this project was to create an application that can, with deep learning, recognize a plant in an image as lupine, hogweeds or impatiens. The application was developed by experimenting with different parameters in the learning algorithm and by observing their effects on results.
The algorithm is based on deep learning techniques. The main language to develop the code was Python. Most of the code is based on PyTorch and fastai libraries and the implementations of artificial intelligence theories. Google Research Colab was used to do the experiments in the cloud. The application was deployed to Heroku.
The application can recognize invasive plants and the success rate of the test results is over ninety percent. Since recognition is the first and vital step to take in order to remove invasive plants from an environment, the results of this project can be applied to robotics engineering in the future.
In conclusion, it is possible to use deep learning in order to clean nature from invasive plants. This means that the next step for this project is to apply the learning model to robots. However, robotics regulations and the ethical aspects concerning the use of robots should be studied first.
The algorithm is based on deep learning techniques. The main language to develop the code was Python. Most of the code is based on PyTorch and fastai libraries and the implementations of artificial intelligence theories. Google Research Colab was used to do the experiments in the cloud. The application was deployed to Heroku.
The application can recognize invasive plants and the success rate of the test results is over ninety percent. Since recognition is the first and vital step to take in order to remove invasive plants from an environment, the results of this project can be applied to robotics engineering in the future.
In conclusion, it is possible to use deep learning in order to clean nature from invasive plants. This means that the next step for this project is to apply the learning model to robots. However, robotics regulations and the ethical aspects concerning the use of robots should be studied first.