Implementation

• Explores existing and proposes new image processing and augmentation procedures aiming to improve small object detection for the drone-based images.
• Evaluates the techniques on EfficientDet, RetinaNet, YOLOv3-SPP, YOLOv5 models.