In training a mushroom vs. non-mushroom image classifier, which images should comprise the negative class?
Only you can answer this question. The answer requires some careful thinking on your part, and it depends on what types of objects that you believe your classifier will encounter once deployed in the "real world". If you are only going to present "vegetation" to your classifier, then you probably only need to curate a
mushroom (positive class) and
plant (negative class) dataset. On the other hand, if there are no limits to what your classifier will encounter, then your training images for the
non-mushroom class must include the
visible universe (minus the mushrooms, of course). Ideally, your positive and negative class images should be collected in the same manner, at least how they would be collected when implemented.