PlantAI: Application for classification and location of plants at risk of extinction in the Atlantic Forest
Abstract
Identification of species of plants in extinction is an important but complex task. The Lista Vermelha project exists to catalog list of plants classified into different threat levels, and has a team composed of professionals responsible for assessing the risk of extinction of species of brazilian flora. However, due to the diversity and similarity contained in Brazilian biomes, the correct identification of plants is not a trivial task. Technological advances have contributed to this goal, making it possible to obtain a large amount of data from different sources, which motivated this work to present a prototype smartphone application instructed to classify plants with different threat levels of the Mata Atlˆantica, and map them collaboratively. A Convolutional Neural Network was trained with augmented samples of different plant species, using the transfer learning technique in different convolution blocks of the MobileNet model, and is used for online classification of image collected with the application. The experiments were conducted in the city of Jacare´ı-SP. For the two species evaluated, Araucaria and Pitanga, it was obtained an accuracy of more than 90%.
References
Degrossi, L. C. (2015). Uma abordagem para obtenção e disponibilização em tempo real de informações geográficas voluntárias no contexto de gestão de risco de inundação. PhD thesis, Universidade de São Paulo.
Flora do Brasil 2020 em construção (2020). Jardim botânico do rio de janeiro. Disponível em: < https://bit.ly/3drneKf >. Acesso em: 2020-03-15.
Goodchild, M. F. (2007). Citizens as sensors: the world of volunteered geography. Geo- Journal, 69(4):211–221.
Goodchild, M. F. and Li, L. (2012). Assuring the quality of volunteered geographic information. Spatial statistics, 1:110–120.
Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning. MIT Press. http://www.deeplearningbook.org.
Hirota, M. M. (2016). A mata atlântica invisível nas cidades. Disponível em: < https: //bit.ly/3afkfma >. Acesso em: 2020-03-13.
Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., and Adam, H. (2017). Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861.
Krizhevsky, A., Sutskever, I., and Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In Advances in neural information processing systems, pages 1097–1105.
Kumar, N., Belhumeur, P. N., Biswas, A., Jacobs, D. W., Kress, W. J., Lopez, I. C., and Soares, J. V. (2012). Leafsnap: A computer vision system for automatic plant species identification. In Computer vision–ECCV 2012, pages 502–516. Springer.
LeCun, Y., Bengio, Y., and Hinton, G. (2015). Deep learning. nature, 521(7553):436– 444.
MMA(1998). Primeiro relatório nacional para a convenção sobre diversidade biológica. https://bit.ly/2UBftZN.
MMA (2016). Quinto Relatório Nacional para a Convenção sobre Diversidade Biológica. https://bit.ly/2UgPR5e.
MMA(2018). Mata atlântica. https://bit.ly/3aePoWS. Acesso em: 2020-03-15.
Nielsen, M. A. (2015). Neural networks and deep learning, volume 2018. Determination press San Francisco, CA, USA:.
Perez, L. and Wang, J. (2017). The effectiveness of data augmentation in image classification using deep learning. arXiv preprint arXiv:1712.04621.
Rawat, W. and Wang, Z. (2017). Deep convolutional neural networks for image classification: A comprehensive review. Neural Computation, 29(9):2352–2449. PMID: 28599112.
Schutt, R. and O’Neil, C. (2013). Doing Data Science: Straight Talk from the Frontline. O’Reilly Media, Inc.
Yosinski, J., Clune, J., Bengio, Y., and Lipson, H. (2014). How transferable are features in deep neural networks? In Advances in neural information processing systems, pages 3320–3328.
