Faster R-CNN: Faster than Fast R-CNN!

Faster R-CNN explanation
Paper link: arxiv.org/abs/1506.01497
Table of Content:
00:00 Fast-RCNN problem
00:33 RPN
04:10 RPN Loss
09:12 Learning procedure
Icon made by Freepik from flaticon.com

[@saeed577]

The best Faster R-CNN explanation video on KZbin 👍👍

[@pranavgandhiprojects]

veryy veryyy well explained..... i also loved your video on fast rcnn:) amazing workk

[@seyedmatintavakoliafshari8272]

Great job Soroush!

[@soroushmehraban]

Thanks :D

[@kelixoderamirez]

permsiion to learn sir. thank you

[@Javad-ek4es]

Nice job! May you please give some link to access your slides, too?

[@mohammadalimohammadkhani5107]

great job, can you explain mask-cnn too?

[@soroushmehraban]

Thanks for the feedback. I’m afraid my field of research is different. I guess I can’t do it as lots of new papers are coming up and I have to focus on them

[@user-ff1qi2yk9q]

thank you so much sir. is the RPN and FAST RCNN trained separately?

[@soroushmehraban]

As I explained from 9:22, the training is done in multiple stages. In the first stage both backbone + RPN are trained. In second stage the backbone is reinitialized and RPN is frozen and network is trained similar to the FAST RCNN. In third stage backbone is frozen and RPN is trained. And in final step both RPN and backbone are frozen and FC layers at the end are trained.