How to solve ssegmentation problem in neural networks?
Answers
Answered by
0
Neutral Network is appropriate for division dependent on individual pixels shading. For example in the event that you wish to choose vegetation district from satellite picture, train the system with physically chosen pixels from test pictures speaking to vegetation and after that utilization prepared NN to order every pixel of the obscure picture.
Such a system will regularly have 3 input neurons (for R, G and B shading segments), couple of neurons in shrouded layer (more neurons gives progressively complex limit between classes, which isn't constantly proper) and yield neurons, which number is equivalent to the quantity of classes. In above precedent, on the off chance that you have to decide whether specific pixel speaks to vegetation or not, one yield neuron is adequate.
Such a system will regularly have 3 input neurons (for R, G and B shading segments), couple of neurons in shrouded layer (more neurons gives progressively complex limit between classes, which isn't constantly proper) and yield neurons, which number is equivalent to the quantity of classes. In above precedent, on the off chance that you have to decide whether specific pixel speaks to vegetation or not, one yield neuron is adequate.
Similar questions