Captur uses machine learning models to train a 'virtual agent' to see what a real agent can see. This is done through a ton of image data and intricate machine-learning methods.
Once Captur receives an image, it will go directly through a model for analysis. The model that it is put through is dependent on the location and asset type. The model will analyse the image and output a decision in under 3 seconds.
As machine learning models go, we can never be 100% right! But we can get as close as possible and that is with the help of the experts (you!) reviewing and giving us the best feedback.