TreeLogy: A Mobile Application for Leaf-Based Tree Identification
TreeLogy is a mobile application that has been designed to perform leaf-based tree identification among tree species in Turkey using only one picture of a given leaf. The app is developed by Duman Emre Akın and offered for free in the Education & Reference category. Unlike other tree identification applications that primarily focus on North American and European tree species, TreeLogy has been localized to support identifying tree species native to Turkey.
The developers of TreeLogy have focused on supervised learning for classification tasks and experimented with both Deep Learning (specifically Deep Convolutional Neural Networks) and Support Vector Machines. The tree identification process uses leaf image features gathered from Caffe, a convolutional neural network framework, and an image processing module. After several experiments, the developers have reached an optimal classification accuracy of 93.59% for 57 tree species using 16,096 training and 3,020 testing leaf images. According to their findings, certain image processing procedures for extracting features such as shape and texture descriptors do not produce features as feasible as convolutional neural networks.
TreeLogy is a much-needed addition to the app market, filling the gap for identifying tree species native to Turkey. The app's use of machine learning and image processing technologies makes it a reliable tool for identifying tree species with high accuracy.
