Klasifikasi Pengemudi yang Terdistraksi Menggunakan Densenet 169

The study aimed to detect distracted drivers using the DenseNet-169 architecture, one of the causes being driver distractions due to road accidents. This can lead to both material losses and loss of lives. The used dataset is the State Farm Distracted Driver Detection, which consists of 19,000 images across 10 different classes. DenseNet-169 was selected for its superior capabilities in dealing with large and complex datasets with high parameter efficiency. The model was trained with pre-trained weights from ImageNet and evaluated with the metrics of accuracy, precision, recall, and F1-score. The findings of the study reveal that DenseNet-169 performs quite well in the classification of the state of distracted drivers with an accuracy of 97.50% on the test data. It is hoped that this technology could contribute to road safety by detecting distracted drivers.