In these days, facial identification has been wide-spread in our daily life. But what about our cars? Can we predict the car brand by building up a model based on car images? We wish this interesting technique may be used for information collections like parking fields or toll station. Also, it may help in detecting car brands in traffic violations. In addition, we may be interested in the specific car brands if we are attracted one fancy car in our daily life. Therefore, this kind of model may be incorporated into car apps and provide relating informations to users after predicting the car brands.
Super resolution/CNN model
The size of each images are quite different.
We incorporate super resolution and cooperate these two models. And the accuracy of the testing data looks good.
First time to use super resolution. Dealing with large amount of data.
We hope we can incorporate more brands into the data and model. Also, we make add more types/colors into our dataset. We believe our next model may precisely predict more information of the cars.