Birdylx
Birdylx
@JaydencoolCC Hi, here is my discovery about the randomness. It seems like the shuffle operation in `sample_points` method over here: [https://github.com/yifanzhang713/IA-SSD/blob/67db5159260474c9afb2e34261e1fc95f56107b5/pcdet/datasets/processor/data_processor.py#L215](https://github.com/yifanzhang713/IA-SSD/blob/67db5159260474c9afb2e34261e1fc95f56107b5/pcdet/datasets/processor/data_processor.py#L215) if you set numpy with fixed seed, you might...
@itskyf thanks for your quick replay！
hi @itskyf , I got a question about the backbone you used in your code. The paper mentioned that you used three `SA layer` and sampled 512 points in Appendix,...
TIPC LGTM
TIPC LGTM
@helloyizhengjun Hi, Paddle3D当前支持nuScenes数据集，是可以直接使用的，为何有转换kitti的需求？
@Skye2099 当前Paddle3D主要聚焦于自动驾驶感知，目前不支持ShapNet数据集，我们会评估下是否加入。当然，你可以参考[这里](https://github.com/PaddlePaddle/Paddle3D/blob/develop/paddle3d/datasets/semantic_kitti/semantic_kitti_seg.py)实现自定义的Dataset
@zyxcambridge Hi, the demo gif is from waymo train set. We just simply draw the gt on point cloud & camera image, you can refer [here](https://github.com/waymo-research/waymo-open-dataset/blob/master/tutorial/tutorial_camera_only.ipynb) for visualization. It's just...
> so the result gif is from gt not predict？ Exactly, it's just a gt visualization gif demo. We will add visualization capabilities in the future, please stay tuned.
same question