The outline is: you create the table on the destination with the same table definition and execute the DISCARD TABLESPACE command on the destination. Execute FLUSH TABLES FOR EXPORT on the source, which ensures that changes to the named table have been flushed to disk, and so a binary table copy can be made while the instance is running. After that statement, the table is locked and does not accept any writes; however, reads can happen. You can copy the .ibd file of that table to the destination, execute UNLOCK tables on source, and finally execute the IMPORT TABLESPACE command, which accepts the copied .ibd file.
For example, you want to copy the events_history table in a test database from one server (source) to another ...