machine-learning - Output of Image recognition as a label in supervised learning

Question

I am new to machine learning and I am stuck in understanding something. I want to create a model in which an Image of a car crash/scratch is recognized and then use that Image to measure the speed of the car. I have a good amount of data with car crash images and the related speed to them.

So basically, I want a model to first recognize the car crash in the image and then recognize the speed related to it. I already have a model which recognizes the car crash using CNN but I don't understand how can I proceed further to predict the speed using the crash.

Please suggest me something. I am pretty new to ML.

标签： machine-learningdeep-learningconv-neural-network