Set 1% FDR (preferably for distinct sequences) and in report builder add a filter 'Num. of significant unique sequences' > 1. This leaves a list of 71 proteins.
Notice how some families are reduced to a single member, e.g. family 12. Do you think one unique peptide is sufficient evidence for HS71A_HUMAN?
Apply the same criteria to the decoy results and there are no proteins.