Currently we are using pertained models downloaded from TF hub. These models work fairly well but have been trained on a pretty generic dataset. Future improvements to this project would benefit from a modified ML model that includes a larger dataset of vehicle or even a custom ML model based on data from installed parking lot locations. Due to time constraints a large dataset could not be acquired during this phase of the project. To that end a ML training repo has been created to help future development on KB4YG.
Currently we are using pertained models downloaded from TF hub. These models work fairly well but have been trained on a pretty generic dataset. Future improvements to this project would benefit from a modified ML model that includes a larger dataset of vehicle or even a custom ML model based on data from installed parking lot locations. Due to time constraints a large dataset could not be acquired during this phase of the project. To that end a ML training repo has been created to help future development on KB4YG.