Off-site construction aims to shift construction work to factory environment wherein enabling modular paneling to be automated. Whereas this approach has proven to be advantageous for the Canadian industry, the changes in the manufacturing processes of panelized walls bring new challenges and opportunities to the construction industry. Within such a controlled environment, inspection for safety and product quality can be automatized. A visual sensor can perform such duties for frame assemblies. This paper proposes a vision-based framework for automatic supervision of the light-gauge steel frame pre-manufacturing stage. The proposed framework is implemented as a Python-based program on the machine environment. The information extracted from an industrial camera placed on top of a steel framing machine prototype is compared with the manufacturing information available from the building information model (BIM) for each frame. The proposed framework is tested on several simulated and real scenarios to validate its accuracy and limitations. The results show this approach successfully identifies, validates and corrects, if needed, the frame assembly on its pre-manufacturing stage.