CODEWITHSUNDEEP
image-processingcomputer-visionsvmencode
wannabegeek
wannabegeek

Reputation: 47

SVM classifier and test images

I have 5 classes in the training set where each class has 100 training images. I have created 5 classes in the test folder. Now each test image will be compared against the training classes and based on it's closeness to one of the training classes it will be labelled for the corresponding test class. I want to know if I am doing the right steps in SVM classifier ?

Upvotes: 0

Views: 445

Answers (1)

Liviu Stefan
Liviu Stefan

Reputation: 183

So basically you have the following structure:

train

|-- class1
|   |-- 1.pgm
|   |-- ...
|   |-- 100.pgm
|-- class2
|   |-- 1.pgm
|   |-- ...
|   |-- 100.pgm
...
|-- class5
|   |-- 1.pgm
|   |-- ...
|   |-- 100.pgm

test

|-- class1
|   |-- 1.pgm
|   |-- ...
|   |-- n.pgm

|-- class5
|   |-- 1.pgm
|   |-- ...
|   |-- n.pgm

The next step is to extract a descriptor from the train images. Using the extracted features, you generate (train) a classification model. In the end, you use the same encoding method to extract the features from the test images and then use the trained model to make predictions.

Upvotes: 1

Related Questions