Is it accurate to say that wavelets were a promising avenue of research in the 1990s but interest in them has kind of died because they were obsoleted by CNNs?