白鹭先生

An Online Demo

11

We have built an online demo demo based on your work at [Click here](https://huggingface.co/spaces/Egrt/Luuu)

对于世界坐标到相机坐标系的校准，首先将包含车道和人行横道端点的高清地图投影到 2D 图像上进行匹配，获得初步变换矩阵，然后通过捆绑调整优化得到最终转换矩阵。 请问这个过程具体是如何实现的，有没有参考方法？

[Feature] rtmpose modified to bottomup mode

1

### What is the feature? How can I change the rtmpose profile to bottomup mode, i.e. make the input image the whole image instead of the cropped target image? ###...

tensorrt的支持

7

请问能否增加对tensorrt的支持？

希望可以增加对YOLO-World的支持

GDLoss

1

When using GDLoss to calculate the loss of 3D target detection, no matter how to modify the angle, it has no effect on the value of loss？ ``` import torch...