EfficientNet is a family of convolutional neural networks and these models efficiently scale up in terms of layer depth, layer width, input resolution, or a combination of all of these factors. EfficientNet allows us to form features from images that can later be passed into a classifier. This allows for EfficientNet to serve as a backbone to many other models--one of which is EfficientDet, an object detection model family. This version of EfficientNEt is implemented in Keras, which is abstracted, so we can load a custom dataset and train EfficientNet all in a few lines of code.
EfficientNet is able to achieve higher accuracies and efficiencies all while reducing the parameter size and FLOPS.
Training EfficientNet: https://blog.roboflow.com/how-to-train-efficientnet/
EfficientNet
is licensed under a
license.
You can use Roboflow Inference to deploy a
EfficientNet
API on your hardware. You can deploy the model on CPU (i.e. Raspberry Pi, AI PCs) and GPU devices (i.e. NVIDIA Jetson, NVIDIA T4).
Below are instructions on how to deploy your own model API.