Drones are used widely in various fields such as security monitoring, traffic control, party photos etc. There are numerous objects present in every image captured by a drone. The detection of objects present in an image captured by a drone can solve real-world problems.
The retinaNet model is present as a separate jupyter notebook file and the 5 files named "model1" to "model5" represent the 5 fasterRCNN models trained for our VISDRONE dataset with each file training 2 distinct classes.
Dataset Link: https://github.com/VisDrone/VisDrone-Dataset