Smart Eyewear: African Lovebirds Classification Using Convolutional Neural Networks

Many people have difficulty classifying animals, specifically African lovebirds, because of similarities between physical aspects. A classification algorithm will be needed to classify the kind of bird from others correctly. In this paper, the researchers created eyewear that can classify four African lovebirds using SVM as color-based feature extraction and CNN as image classification. The researchers will only focus on identifying Green Personata, Green Fischer, Nyasa, and Black Cheek lovebirds, whereas other kinds are not included in the study. Sixteen thousand images were gathered to train the classification models to classify African lovebirds accurately. Each African lovebird has a maximum of 2,400 images, which will be allotted for testing the created classification model. The results showed that the eyewear could classify four African lovebirds, validated using a confusion matrix. Each African lovebird's accuracy was calculated, leading to an accuracy rate of 85.18%. Overall, the eyewear created will help people identify the type of African lovebird. The output classification will be updated in real time, which can also help people get used to classifying African lovebirds.