I'm working on project for the Autonomous vehicle, and i want to know what's confidence level means and how can we use confidence level for vehicle detection in OpenCV ?

  • $\begingroup$ The c++ tag has nothing to do with your question. For confidence level, you need to relate your question to robotics otherwise this question is more suitable in Cross Validated Stack stats.stackexchange.com. $\endgroup$
    – CroCo
    Commented Sep 30, 2015 at 23:55
  • 1
    $\begingroup$ Or check Wikipedia for more information about Confidence Intervals. If you really want to use OpenCV for vehicle detection then check OpenCV Q&A to find for example: Detecting an incoming bus in opencv?. $\endgroup$
    – agold
    Commented Oct 1, 2015 at 18:41

1 Answer 1


Statistical confidence is most often defined in terms of a score that is the number of standard deviations away from the mean.

Now let's say you have a set of states (a state vector) used to define an object recognized using image processing in OpenCV (for example, RGB data, HSV data, filter responses, etc.). Given a series of training images for vehicles, you can compute the mean and covariance of their state vectors. If you then consider some new object being tested, you can compute what is known as the Mahalanobis distance between the new object's state vector and the vehicle state vector. This distance will essentially be the statistical confidence that the new object is a vehicle.

In fact, OpenCV has functions available to compute things like the covariance matrix and the Mahalanobis distance.

  • $\begingroup$ Note that this is only one example of how to classify an object based on statistical confidence. There are many algorithms that have a probabilistic basis. For example, you may also want to consider using a Bayesian network. $\endgroup$ Commented Oct 16, 2015 at 9:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.