search

xtreme1-io / xtreme1

Xtreme1 is an all-in-one data labeling and annotation platform for multimodal data training and supports 3D LiDAR point cloud, image, and LLM.
https://www.basic.ai
Apache License 2.0
798 stars 126 forks source link

Annotation on accumulation #225

Open ClementLeBihan opened 3 months ago

ClementLeBihan commented 3 months ago

Hi all,

I'm creating this ticket to discuss about an accumulation feature to help annotate LiDAR data.

To accumulate data, we need to know the lidar pose in a global frame. I think we could use the lidar_config to register the lidar pose.

We could then load multiple pointclouds in their corresponding frame, and directly annotate multi-frames.

What do you think about this feature ?

jotamotk commented 3 months ago

This is an absolutely innovative feature. We are still developing it for our cloud version, and it will be coming soon.

ClementLeBihan commented 3 months ago

I did a first try with a setting Add Accumulation that load additional frames'pc and append it in the existing pc ...

Screenshot from 2024-03-26 09-28-37

To do so, I load the lidar_config pose matrix in IFrame lidarExternalPose of each additional frames, and I transform the pointcloud of additional frames in the current frame ... I saw there is a worldMatrix in IFrame object, it may be better to use it, I do not know exactlly what is the better way. Maybe using the worldMatrix of objects can be easier to display multiple frames ... What do you think ?

guhaomine commented 3 months ago

You are right, using worldMatrix can easily operate each point cloud. But integrating such functions into tools is not easy, and we are also trying to complete it.