Reason: the human brain can imagine what it would look like when it turns around given a picture of an unknown object. But machine learning can not do. ML can only successfully classify this given picture or extremely similar pictures.

Solution: combine image generation task, think of it as the human brain’s imagination, and generate as many photos as possible from different angles when classifying a picture. If you classify them, you won’t just use the original photos as the only benchmark.

--

--

Vision is the most important perception.

With sound and bralliant Computer Vision that much much better than human brains, with sound and bralliant telescope that much much better than human eyes, then, nothing is impossible.

--

--