Dear Sir:
I want to use CVB to classify the different Label.
But the label is different size. but Polimago Classification and Regression Project use the same window size, so that i don’t know how to create the classifier.
Requirement:
I have 8 kind different label, each two labels will post on the product.
The AOI Logic will inspect is correct labels on the product base the recipe.
SA:
If I can use CVB to build the label classifier .
You could create a scale invariant search classifier. Then you can locate the labels and get a scale factor. With that factor you could then scale the image (or ROI of an image) accordingly and then classify it. For the classification training set, you would then add all labels at the same reference size. The classification itself does not need to be scale invariant in that case.
Is I need to scale my image for different label size and save to be image data-set for PolimagoClassification Project?
Is it use image pool tool to resize image for different label ?
and the aspect ratio of the label is different…
I don’t understand how to create a scale invariant search classifier.
I’d recommend to scale all labels to a certain reference size for the classification training set.
In regard to the aspect ration, you could use just a subsection of the label for classification. Maybe you will be able to identify certain sections of the label that enable you to determine the class just based on that subsection alone; You might not need the complete label to come to a satisfying classification conclusion.
If you want to create a scale invariant search classifiers you also have to size all instances/images, that make up your training set, to a certain reference size. To then create a scale invariant classifier you have to use the ScaleRotationTranslation invariance type and set the scale range according to your use case:
