I'm attempting to train a new data set using the Mobilenet_v1_.5_128 pretrained network. I have a few questions.
1) based on the tutorial here https://pythonprogramming.net/introduction-use-tensorflow-object-detection-api-tutorial/ which I've already followed, I need a .config file for the training. Where do I find one that will work with Mobilenet_v1_.5_128 which I found at this link https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet_v1.md. It does not come with a .config file that I can modify to point at my tensorflow records, etc
In the tutorial above, the .config file came from this repo https://github.com/tensorflow/models/tree/master/research/object_detection/samples/configs which has a number of .config files. I'm assuming that the .config file should line up with the network I'm using.
2)Image size, can I use higher resolution images? for instance, I have images that are 1280x720. I understand training will take longer, but is it still possible to train with this image size?
For more complete information about compiler optimizations, see our Optimization Notice.