2D detector

[stomachacheGE]

Hi Charles,

Thanks for your excellent work and sharing source code with us.

Since the performance of F-pointnet is highly dependent on the 2D-detector, I am curious whether you will release the 2D detector you trained for this work. Although I can integrate other 2D detector into your framework, the overall performance is not as satisfactory as I expected. I will be very appreciated if you can share the 2D detector with us :)

Best regards

Liangcheng

[bkanchan]

Hi Charles.

I agree with @stomachacheGE , I will truely appreciate if you can share the 2D detector with us. That will help me to test the framework with additional data.

Thanks, Kanchan

[PranjalBiswas]

Hi Charles,

Firstly, I would like to appreciate you for the great work done. Also, as mentioned above by fellow community members it would be really nice if the code for 2D object detection algorithm can be released. If it will take time, then in the meanwhiile can you suggest which 2D detector will be the best suited for your algorithm, I was planning on to use YOLO v3 so that I can proceed with my project. Hope to hear from as soon as possible.

Kind Regards Pranjal Biswas

Update: I ended up using YOLOv3 and it works fine along with FPN (not real time, neither where results in FPN paper validated using YOLOv3. The pipeline stores results from YOLOv3 in memory which is then loaded in by FPN.).

[Z0org]

Hi, does someone now which model for the 2D detection is used by now? @PranjaLBiswas27 do you get the same results as stated in the paper with YOLOv3?

[jenbu]

@PranjaLBiswas27

Hi Charles,

Firstly, I would like to appreciate you for the great work done. Also, as mentioned above by fellow community members it would be really nice if the code for 2D object detection algorithm can be released. If it will take time, then in the meanwhiile can you suggest which 2D detector will be the best suited for your algorithm, I was planning on to use YOLO v3 so that I can proceed with my project. Hope to hear from as soon as possible.

Kind Regards Pranjal Biswas

Update: I ended up using YOLOv3 and it works fine along with FPN.

How did it run in realtime with YOLOv3?

[shengrongjin]

@PranjaLBiswas27 Hi PranjaLBiswas27， Please, I'd like to ask you Which one 2D object is more suitable , YOLO v3 or MASK-RCNN ?

[lhyfst]

Thank you Charles for your great work! I will appreciate it as well if you could open source your 2d detection network. For another thing, I think the 2D proposals should be generated in real time, and this time should be added to the sum time of evaluation, or it will be unfair for other methods when we compare their speeds. Thanks

[chengm0-0]

Hi, I want to know how to combine my 2D-detector's output with your code? Thanks!

[cheon12]

hi charles

[yutang814]

@chengm0-0 Hi,chengm0-0 Did you get any answer? I have the same question.Thanks!