i made a cnn and from the third conv layer i extracted the features so that the shape is (600,12,12,128)
600 = training images 12x12128 = for a single image
No