Abstract:
In this technological era, computer vision accomplishes superior performance and it
introduced to the Convolutional neural network (CNN) which is acquainted for recognize
an object. In addition to object detection and classification are considered difficult tasks
in computer vision. From the ancient times vehicle is the only communication media to
move from one place to another place. In this automation period local vehicles are
decreasing day by day. Because these type of vehicles motion are slow. We want to
preserve our culture with the help of vehicle-NN model.We presented our own CNN
model name Vehicle-NN which will be a convenient effect on the vehicle sector of
Bangladesh. We compared our CNN model with other pre-trained models like
MobileNet, VGG16, InceptionV3. This proposed model plays an effective role in the
future automation vehicle identification sector. Besides we want to hold on to our
traditional roots and preserving our local culture. Our work approach is novel for the
detection of local vehicles in Bangladesh. In this experiment, we have used 7 classes
(Bus, CNG, Leguna, Pickup, Rickshaw, Thelagari, Van). We trained our model with our
own dataset and our CNN model acquired accuracy is 96%.
