longyunf / radiant

21 stars 3 forks source link

How to caculate mAAE in val and test? #16

Open HoHoHO1999 opened 3 months ago

HoHoHO1999 commented 3 months ago

Hello, I wan to ask how to caculate mAAE in val and test? the value always be 1.000,thank you

longyunf commented 3 months ago

As the fusion only updates positions, you can keep other parameters from monocular detections. mAAE should be almost the same as monocular mAAE.

HoHoHO1999 commented 3 months ago

As the fusion only updates positions, you can keep other parameters from monocular detections. mAAE should be almost the same as monocular mAAE.

Hello, because the scores displayed through nuscenes are all 1.000, I don’t know if there is a way to get the correct scores. I also want to ask how I can visualize the result.json generated by do_eval. I don’t know. Why are the visualized results so strange, as shown below? Can you provide the relevant program? thank you

23dc1b528c734082b8b3bfd9757f0eb3

QilongWu2 commented 2 months ago

由于融合仅更新位置,因此您可以保留单目检测的其他参数。mAAE 应该与单目 mAAE 几乎相同。

你好,因为通过nuscenes显示的分数都是1.000,不知道有没有办法得到正确的分数,另外想问一下如何可视化do_eval生成的result.json,不知道,为什么可视化结果这么奇怪,如下图,能提供相关程序吗,谢谢

23dc1b528c734082b8b3bfd9757f0eb3

As the fusion only updates positions, you can keep other parameters from monocular detections. mAAE should be almost the same as monocular mAAE.

Hello, because the scores displayed through nuscenes are all 1.000, I don’t know if there is a way to get the correct scores. I also want to ask how I can visualize the result.json generated by do_eval. I don’t know. Why are the visualized results so strange, as shown below? Can you provide the relevant program? thank you

23dc1b528c734082b8b3bfd9757f0eb3

hello,Have you implemented the visualization function? I'm going to need visualization now, too, and I thought we could talk about it, and we look forward to hearing from you