Superior Performance in Object Detection and Localization. Top performance in Classification. Is this for you?
You really should put in the training cost/time for there architectural search. This might be useful at Google but not for most people doing real application unless they just use Spinenet as a 'better' ResNet.
Great call. I wanted to focus on the idea of spine permutation, but you're spot on that this will not be realistic for most organizations.
You really should put in the training cost/time for there architectural search. This might be useful at Google but not for most people doing real application unless they just use Spinenet as a 'better' ResNet.
Great call. I wanted to focus on the idea of spine permutation, but you're spot on that this will not be realistic for most organizations.