EfficientDets are a family of object detection models, which achieve state-of-the-art accuracy, yet being an order-of-magnitude smaller and more efficient than previous models. EfficientDets are developed based on the advanced backbone, a new BiFPN, and a new scaling technique: • Backbone: we employ the more advanced EfficientNets as our backbone networks. • BiFPN: we propose a new bi-directional feature network, named • BiFPN, to enable easy and fast feature fusion. In addition to the bi-directional topology, we also propose a new fast normalized fusion that enables better fusion with negligible latency cost. • Scaling: we propose to use a single compound scaling factor to govern the network depth, width, and resolution for all backbone, feature network, and prediction networks.
Don't forget to add the tag @toandaominh1997 in your comments.