extract features from low resolution

Question

extract features from low resolution

samo

2022年5月23日 20:21

I have medical images and need to extract features from the layer before the classification layer using VGG for example but the resolution of the images is not efficient... Are the features without improving this resolution will not be affected or do I need to improve the resolution before extracting the features?

I was doing processing in color images for extracting the features using VGG by this processing

preprocess = T.Compose([
    T.Resize(256, interpolation=3),
    T.CenterCrop(224),
    T.ToTensor(),
    T.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),
])
image= load_img(aa.jpg, target_size=(224, 224,3))
proc=preprocess(image)

what if the images I have are grayscale or blur will this processing be suitable for them or do I need to change?

Topic vgg16 deep-learning feature-extraction

Category Data Science

lpounng · Accepted Answer · 2022年5月23日 10:15

IIUC, the question is "would blur images affect the features extracted from VGG?"

Yes and no, depends on your application (i.e. "what do I want to use these features for?"). In general we don't know till we try. Rule of thumb is that if a human doctor can do it, there is at least a chance to work.

But technically, there shouldn't be anything stopping you from training/fine-tuning an VGG architecture and extract the features from it. Note that you will need to adapt to the input size (e.g. 224x224x3 for VGG16) either by transforming your images or replacing the input layer of model.

Edit (to OP's additional comment)

Looks fine to me (though I admit not knowing what the interpolation does). It is a common practice to populate the 3 color channels with grayscale. There may be more advance techniques but I cannot recall (quite a while since I last work on medical image).

extract features from low resolution

About