The use of the Convolutional Neural Network (CNN) model for the classification of bird species is the focus of this study. The methodology used in this study utilizes the BIRDS 400 - SPECIES IMAGE CLASSIFICATION dataset to be tested with the CNN InceptionV3 model and is limited to only 18 classes. The experiment was run on the free version of the Google Collabs platform. From the results of experiments conducted, the CNN InceptionV3 model requires time to train images for 36 minutes 46 seconds and obtains a training accuracy of 93.36%. For experiments at the testing stage, the CNN InceptionV3 model that has been trained on the Google Collabs platform can correctly predict all images in the testing folder. Based on the results of the research conducted, the CNN InceptionV3 model has very high accuracy for handling the dataset and the free version of the Google Collabs platform is sufficiently capable of processing the dataset.
Copyrights © 2023