1
$\begingroup$

I understand that CNNs are for image classification while object detection is for localization + classification of the objects detected. However, in particular, AI for chest radiographs, why is object detection used? If a CNN has 99% accuracy, should object detection still be considered? I see a lot of research papers on object detection with x-ray data but they don't explain why object detection is better than CNNs. While object detection allows users to see "where" the object is located, does this even matter if we get such high accuracy already? Also, if the location really does matter, can't we just use the heat maps from CNN?

$\endgroup$

1 Answer 1

1
$\begingroup$

I think you might have misunderstood 2 concepts here: CNNs and Object Detection.

Object Detection is an AI approach to solve problems where you are interested in both the location and the classification of key elements in the image. On the other hand Image Classification is another approach where you are interested in classify the whole image with a tag.

Those are very well known approaches to solve computer vision problems through AI. There lots of approaches, you select the one that outputs the information you are most interested in.

CNNs is network type. You can build Object Detectors or Image Classifiers with CNNs, but you can also build it with Transformers, with Multi Layer Perceptrons, with LSTM, even there are some approaches with Reinforcement Learning.

Going back to your problem of AI for chest radiographs, when you see 99% accuracy it is probably in Image Classification (predict probability of bone broken in the image). On the other hand Object Detection is a more informative approach because it locates the places where a bone could be broken and the probability of that bone being broken. It is more informative because the doctor now knows where to look in the chest radio image.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .