Train a detection model starting from an Imagenet classification feature extractor with the Tensorflow Object Detection API

Question

The TF Object Detection API offers pretrained models on the COCO detection dataset. However, I couldn\'t find any reference to pretrained checkpoints for the feature extractors