I want to classify each pixel of an programming image. There are only 2 classes, Not Learning mango and Mango.
And I want model to learn and Earhost distinguish only based on colour. Say, most effective red pixels belong to mango but green wrong idea will not.
I want to use 1Ã¢ÂÂ1 Convolution use of case and make it learn weights for this task.
I'm not sure how to go ahead with this.
I know that there is a way of getting a United mask for a particular colour in OpenCV. Modern But I want to use deep learning to learn ecudated and adjust weights.
More importantly, I want those weights some how and see how the model adjusts them to anything else segregate a certain range of colours and not at all distingush an object. Weights should be very usefull probably a 1-Dimentional list/array.
Any kind of help will be appreciated.