The objective of the work is to test the idea of training a machine learning model using boundary conditions as an input to predict flow fields in arbitrary incompressible viscous flows. Currently, the flow field in the standard lid-driven cavity at any value of Reynold number (Re) is predicted using the data-driven approach, specifically by implementing deep neural networks. The boundary conditions are used as an input layer for training the ML model. Standard computational fluid dynamics (CFD) simulations were performed using commercially available CFD solver ANSYS (Fluent) over a range of Reynolds numbers and validated with the available literature data before training the machine learning algorithm. Subsequently, the results obtained from the boundary condition-based machine learning model (BCML) are compared with those obtained using the CFD data. The current study is a preliminary attempt to understand and will be extended to general 2D incompressible viscous flows.