# The computation of the predicted probability in a cell

Hi all,

I was trying to get some clue on how to compute the probability (Pc) of object in a given cell.

Let’s say we want to compute the probability of detecting a car, how to proceed? I am just curious about how to get these probabilities.

I thought it would be something like (# occurrences of a particular class) / #classes. the number of classes (#classes) is known in advance, so the problem remains on the numerator if it the above formula that is used. When the front camera snaps something and send it to the system in real time, how does the system computes the probability?

Thank you!

At training time, the probability for a given location is 1.0 if an object is present and 0.0 if there is not. The output of the CNN includes a value for object presence. The predicted value is compared to the ground truth value in the loss function, and hopefully iteratively approaches it.

At runtime, the CNN outputs that object presence value given the input signal and its learned weights. The value is constrained to be between 0.0 \leq x \leq 1.0 so that it can be treated as a confidence or probability. [know any activation function that would be helpful???]

There is an example of setting up the training data that touches on this in one of the replies in this Discourse thread: Quick question regarding YOLO algorithm - #3 by ai_curious

1 Like