Texture-Based Dipterocarpaceae Trunk Classification Using Two Stage Transfer Learning of VGG16

The significance of plant identification has already been recognized by academia and industry. There were several attempts to utilize leaves and flowers for identification. However, the trunk also could be beneficial, especially for tall trees. This research focuses on the Dipterocarpaceae family classification that could attain heights between 70 and 85 meters. Leveraging deep learning using VGG16 architecture, this study proposes a two-stage transfer learning strategy. In the first stage, the pre-trained VGG16 is fine-tuned by modifying only the classification layer while keeping the feature layer frozen. The second stage involves unfreezing the last three convolutional layers to adapt the model more specifically to Dipterocarpaceae trunk classification. A dataset containing 681 images of different Dipterocarp species was collected. Through experimentation, the proposed method achieves a remarkable accuracy of 97.09% on the testing set, outperforming traditional transfer learning method. This research highlights the approach's potential for accurate species identification, offering benefits to conservation and ecological studies by enabling rapid and reliable classification of tree species based on trunk texture images.