This paper discusses the application of three pre trained Road debris is a serious problem that can cause This paper discusses the application of three pre trained Road debris is a serious problem that can cause accidents on roads, especially on highways where vehicles are traveling at high speeds. This debris can be in the form of substances, materials, or objects that are foreign to normal roads. Common types of road debris include barrels, car parts, puddles, salts, and trees. Detecting and classifying road debris can be challenging, but recent advancements in deep learning have made it possible to develop accurate and efficient models for this task. In this paper, three pretrained convolutional neural network (CNN) models were applied to classify five different types of road debris. The models used in the study were VGG16, MobileNetV2, and InceptionResNetV2. The dataset used in the models consisted of images of the five different types of debris, which were downloaded from the internet. The dataset was divided into three sets: training, validation, and testing. In the first phase, the dataset was divided into sets with 146, 73, and 49 images, respectively. In the second phase, after image augmentation, the dataset was divided into 875 training images, 375 validation images, and 114 testing images. The performance of the models was evaluated over 10, 20, and 30 epochs. The learning rate was set to 0.0001 with an Adam optimizer and a batch size of 10. After image augmentation, the VGG16 model achieved the best performance with a training accuracy of 100% and a validation accuracy of 96.65%. The VGG16 model was able to correctly classify 103 out of 114 images, resulting in an accuracy of 90.35%. The use of deep learning models for this task has the potential to significantly reduce the number of accidents caused by road debris, thereby improving road safety for all users.