Single particle cryoelectron microscopy (cryoEM) and cryoelectron tomography (cryoET) are powerful methods for unveiling unique and functionally relevant structural states. Aided by mass spectrometry and machine learning, they promise to facilitate the visual exploration of proteomes. Leveraging visual proteomics, we interrogate structures isolated from a complex cellular milieu by cryoEM to identify and classify molecular structures and complexes de novo. By comparing three automated model building programs, CryoID, DeepTracer, and ModelAngelo, we determine the identity of six distinct oligomeric protein complexes from partially purified extracts of the nitrogen-fixing bacterium Azotobacter vinelandii using both anaerobic and aerobic cryoEM, including two original oligomeric structures. Overall, by allowing the study of near-native oligomeric protein states, cryoEM-enabled visual proteomics reveals unique structures that correspond to relevant species observed in situ.