flownet3d
flownet3d copied to clipboard
xingyul
FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019)
Hi, I'd like to use FlowNet3D for motion segmentation. It was described in Section F of [supplemental](https://openaccess.thecvf.com/content_CVPR_2019/supplemental/Liu_FlowNet3D_Learning_Scene_CVPR_2019_supplemental.pdf), however, I couldn't find the implementation of this in the repo. Could you...
Hello, I want to ask the function of optical flow data. In the literature, after input the pointcloud data into flownet3d, it can output the sceneflow. Therefore if we only...
Hi, Thank your for sharing your codes! I'm finetuning the model with KITTI data with ground points. I use the data you provide in [meteornet](https://github.com/xingyul/meteornet). But I cannnot get the...
@xingyul Hello, thanks for your awesome work! However, i find that the implement of upconv. part is different between repo. and paper. In the paper, 1. use different radius to...
Hi, In your paper, you mention that a lambda scaling factor was used to scale the scene flow output when concatenating it with the original xyz values. What was the...
Hi, I have a query regarding the finetuning on KITTI. Was it done with 2048 points as in your FlyingThings3D training script or with 16384 points as with your KITTI...
Hi, For KITTI evaluation, you've stated: "Note that the model used for evaluation is in `model_concat_upsa_eval_kitti.py` instead of the model used for training. The average 3D EPE result is approximately...
Hi, It seems the using the preprocessed data and KITTI evaluation scripts you included a few days ago, I am not getting the same values as in the paper. Am...
Can you share the origin data with ground LiDAR info?