how to do transfer learning for the detector model provided by the Automated Driving toolbox to suite to my need?

I am trying to utilize the detector (uses Computational neural network) provided by automated driving toolbox to detect vehicle from an image generated by a virtual environment generator. The detector works good for high resolution images but fails to detect the vehicle in low resolution images. I chose the low resolution images to improve the processing speed. I was hoping that i could do a transfer learning for the detector so that the vehicle can be detected in such low resolution as well but don't know if that will work or not. Looking for an expert opinion. If it works, how can i achieve it?