Vehicle detection and classification are crucial for traffic analysis. Moreover, accurate and effective traffic flow monitoring has faced numerous challenges and is a task that is only getting harder. It can be difficult to identify and locate cars in images and video frames that contain car features because of interference and the proximity of the cars on the road. Furthermore, no datasets are available for local vehicles in the Philippines. The researchers made a customized dataset for the local cars in the Philippines. The proponents employed Convolutional Neural Network (CNN), specifically the YOLOv4 Model model, for vehicle detection and classification. The researchers used Transfer Learning, which means that the researchers used a pre-trained YOLOv4 model. Moreover, the researchers used Supervised Learning by labeling the datasets through annotations. The researchers used seven augmentation techniques: grayscale, horizontal flip, motion blur, random rotation, color augmentation (hue, brightness, contrast, saturation), random shearing, and Gaussian noise. To increase the mean average precision (mAP) further, the proponents fine-tuned the YOLOv4 model. The researchers tested the hyperparameters for the learning rate and the Batch Size of the YOLOv4 Model. The researchers tested three different learning rates, which are 0.01, 0.001, and 0.0001. On the other hand. the researchers tested three different batch sizes, which are 16, 32, and 64. The results of the study show that the best configuration is a learning rate of 0.001 and a batch size of 64. This configuration for the YOLOv4 Model resulted in an accuracy of 98.83%. Nonetheless, theresearchers have recommended that future researchers develop the model with more datasets and increase the diversity of the images.