1
$\begingroup$

What i mean by the title is as follows:

Say I went into Tensorflows Object Detection Model zoo and picked some model like, Faster R-CNN ResNet101 V1 1024x1024.

My question is, what is the architecture and what is the technique? Does Faster-RCNN point to the technique the architecture (i.e. ResNet 101) uses or is FasterRCNN also a standalone architecture?

If what i'm saying above doesn't make much sense then what do both "Faster RCNN" & "ResNet 101" refer to within the context of the whole name of the model "Faster R-CNN ResNet101 V1 1024x1024"?

Any help would be appreciated.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
2
$\begingroup$

Typically, object detection networks can be broken down into two parts: a backbone network which takes an input image and outputs a feature map, and the "head(s)", which produces the final detections from the feature map.

Backbones can often be swapped out for one another, and they're not usually the focus of research -- there's just much more interesting things to be explored in the heads. So in this context, "Faster R-CNN" refers to the design of the heads, and "ResNet 101" is the name of the backbone network.

Improve this answer
$\endgroup$
1
  • $\begingroup$ So in the case of "RCNN", since the selective search process happens before features are extracted, would you still refer to it as the head of the operations? (or tail maybe would be more appropriate?) $\endgroup$
    – Kamal Raydan
    Commented Jan 8, 2021 at 13:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.