finding all instances of an object in an image

[Asif Shakeel]

Can Caffe be trained to find all the instances of an object in a scene? Or one needs to  trained detector for single object and then scan the image?

Thanks

[Axel Angel]

The problem you have in mind is the "localisation" and "detection" task which is much harder than the "classification" task. There are multiple ways: training a FCNN, use segmentation then process it; or training a classifier CNN and find a way to detect interesting spot "region of interest" or try all patches and classify it. That's all I know, anyone should feel free to correct what I said.

Keywords: sliding window, object detection, object localisation, segmentation, fully convolutional neural network.

[Evan Shelhamer]

Fast R-CNN is a package by Ross Girshick for detection by proposals + convnets that makes use of Caffe. I suggest checking it out.

One should also be aware of Faster R-CNN but note that the code is not yet released.

Evan Shelhamer

--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users...@googlegroups.com.
To post to this group, send email to caffe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/caffe-users/1e652c95-8ee4-4c96-b032-a8d160765974%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[el]

.....

One should also be aware of Faster R-CNN but note that the code is not yet released.

--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users...@googlegroups.com.
To post to this group, send email to caffe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/caffe-users/1e652c95-8ee4-4c96-b032-a8d160765974%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

Code has been released: https://github.com/ShaoqingRen/faster_rcnn