I was thinking about the small object detection problem, and while surfing found a reasonable amount of paper those are working on Faster RCNN to solve this issue. Why they choose Faster RCNN instead of other state-of-the-art methods? I want to know the technical reasons. Thanks in advance.
we face the same problems when detecting cardiovascular pathologies (cinema-MRI segmentation). There is no real standard to recommend. It all depends on the syntax in use.
https://towardsdatascience.com/faster-r-cnn-for-object-detection-a-technical-summary-474c5b857b46
Ramzi Mahmoudi, thanks for your answer. My intension was to know the significant reason behind using FRCNN instead of other SOTA methods. Nowadays there are other SOTA methods (accuracy higher than FRCNN) that are used to solve the SOD problems. Then why still the researcher use FRCNN? Ofcourse there should be a strong reason to do that. could u please share your experiences?
https://www.hindawi.com/journals/jece/2020/3189691/
https://paperswithcode.com/task/small-object-detection/codeless
Maybe this article will be helpful for you.
Rashidul Hasan Hridoy, Thanks for your links. Though I have read that paper earlier, I didn't find any precise answer or future recommendation.
Mehnaz Tabassum Welcome, I will inform you if I get the answer to your question.
I'd go for scaled YOLO-p7, implementation is easy and works well
https://paperswithcode.com/sota/object-detection-on-coco
I am about to write a fully automated trianing notebook as well as a corresponding medium article. Can share with you if you want :)
Christoph Haring, Thank you for your reply. I will happy to read your article.
Hi Mehnaz Tabassum ,
I recommend to check this paper:
Article Efficient Real-Time Human Detection Using Unmanned Aerial Ve...
We achieved to detect people in 7x18 pixel size in a low consumption GPU.
Christoph Haring Could you pls share your article on Medium. I didn't find it there. Thanks.
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning