search

Castiel-Lee / robustness_pc_detector

We propose the first robustness benchmark of point cloud detectors against common corruption patterns. We first introduce different corruption patterns collected for this benchmark and dataset. Then we propose the evaluation metrics used in our benchmark. Finally, we introduce the subject object detection methods and robustness enhancment methods selected for this benchmark.
5 stars 3 forks source link

Reasons for the rain/snow severity configuration #1

Closed boyang9602 closed 2 months ago

boyang9602 commented 2 months ago

Hi,

Thank you for your great work. I am doing some similar work. I wonder what the reasons are for the choices severity of rain and snow.

rain in {5.0, 15.0, 50.0, 150.0, 500.0} and snow in {0.5, 1.5, 5.0, 15.0, 50.0}.

Regards, Bo Yang

Castiel-Lee commented 2 months ago

Hello Bo,

The parameters of rain are based on the investigation of rainfall rate in mm/hr. So do snow parameters. Also, such settings optimize the simulation to real-world scenarios as depicted in the paper.

I hope it helps.

On Mon, Aug 19, 2024 at 4:45 PM boyang9602 @.***> wrote:

Hi,

Thank you for your great work. I am doing some similar work. I wonder what the reasons are for the choices severity of rain and snow.

rain in {5.0, 15.0, 50.0, 150.0, 500.0} and snow in {0.5, 1.5, 5.0, 15.0, 50.0}.

Regards, Bo Yang

— Reply to this email directly, view it on GitHub https://github.com/Castiel-Lee/robustness_pc_detector/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/AVYS7FVQUOZZH4KPOE7ZOHTZSJYQTAVCNFSM6AAAAABMYWTYCCVHI2DSMVQWIX3LMV43ASLTON2WKOZSGQ3TIMZXHEYDCNI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

boyang9602 commented 2 months ago

Hi Castiel,

Thank you for your quick response. I overlooked the paragraph about the similarity between simulated and real point clouds.

I thought that 50 mm/hr was already heavy enough in the real world but I overlooked the gap between the simulation and reality. Thank you again!

Regards, Bo Yang