I had a positive experience with ICEVISION and object detection a year ago.
Going thru the new FASTAI docs and API and seeing BBox Blocks under DataBlock API, I wanted to ask if there is a native solution/example/tutorial to object detection phenomena? or should I refine my knowledge on ICEVISION because that is the only existing high level API for object detection?
Also, I am planning to dig more into mid level API.
‘Principally/notionally’ is midlevel API sufficient to develop a custom model for object detection task?
And in connection to that, would fastai native tutorials (I am talking about the examples under tutorials/intermediate on docs.fast.ai) for midlevel be sufficient for hands-on experience on midlevel API? (well those appear to be the only ones, not counting ‘from scratch’ examples from the book). Again the focus is more on the API, so more on the practical side of the midlevel…
PS: I hope someone brings up ‘object detection’ in part II course, so that we have an update to 2018 object detection sessions.