If the attacker has access to the entire TCP session, they will have sufficient clues to know that the transport protocol is ssh. Try to telnet to port 22 on a server you know is running ssh on a standard port to see how.
If the attacker does traffic analysis, he or she will quite possibly be able to distinguish ssh terminal traffic from ssh port forwarding traffic (though not able to identify the contents directly). Basically, typing does not look like downloading HTML, downloading, or transferring data at any speed / with little latency.
If the attacker has access to all of the network traffic for one of the endpoints, he or she will quite possibly be able to identify what specific fraction of the traffic passing through that endpoint also passed through the tunnel... but you've already lost a lot by that point; not much privacy to preserve.
There isn't really that much information in the clear though. I doubt there is a serious risk here, but without knowing your concerns, it is hard to address them.