Pretrained CNN architecture

I am asked to design the deep learning model for object detection in self-driving car using pre-trained CNN architecture(eg :VGG16,ResNET,MobileNet) for transfer learning and fine tune the selected model on given dataset for object detection. Would you please help me guide the process,Please ?

See the Deep Learning Specialization, course 4 on convolutional neural networks.

And also Tensorflow Advanced Techniques Specialization, if I remember correctly course 3!