This repository provides a PyTorch implementation of the EfficientDet architecture from the 2019 paper by Google Research. EfficientDet is a family of scalable and efficient object detection models using weighted bi-directional feature pyramid networks and compound scaling. The implementation supports EfficientDet-D0 through D7 variants, training on PASCAL VOC and COCO datasets, with features including Focal Loss, ROI Align, and multi-scale training.