Ship Classification and Localisation
Professor : Antonello Rizzi
Supervisor : Emanuele Ferrandino
authors : Danial Zendehdel , Peyman Javaheri.
The aim of this project was to classify ships in 5 different categories such as Cargo , Cruise , Military,Carrier , Tanker and in addition to that find the centriod of each found ship in the image.
The Dataset has been downloaded from Game of Deep Learning website.The link for downloadinf the Dataset is here,then from these images the new CSV file extracted with benefit of using LabelImg with annotations in addition to labels which are included in this repository.
There are 2125 images from Cargo , 916 from Carrier , 776 from Cruise ,1148 from Military and 1217 Tanker.
Two models have been used in here :
Multiple ouput : this is MobileNetV2 base model which is used to get predictions for both bounding boxes and labels classification , some layers added for Regression leg.
Second Model : This one consists two model one Xception exclusive for Classification and MobileNetV2 for regression.
The Weights for Demo can be downloaded from here.
Link for Test Data here