It depends on what group of fishes you want to identify. It is true that barcoding (i.e. COI primers) are the widely used. But you can also find papers that use other markers to identify certain fishes, such as canned tuna or sardines. I suggest you to start with Ward (2005). "DNA barcoding Australia’s fish species". You'll be able to cover most of the marine fishes.
Is there a chance to identify fish species as many as possible ( to not to distinguish their species but to identify that sample is derived from fish) using few sets of primers?