Meanings of detection result

[russellyq]

Hi thanks for your work.

Can you specify the meaning detection result in txt files.

Thanks

[wangxiyang2022]

For 3D detections 0,2,298.3125,165.1800,458.2292,293.4391,8.2981,1.9605,1.8137,4.7549,-4.5720,1.8435,13.5308,-2.1125,-1.7867 which respectively mean frame type(2 means car) 2D BBOXz(x1,y1,x2,y2) score 3D BBOX(h,w,l,x,y,z,rot_y) alpha.

For 2D detections 0,296.021000,160.173000,452.297000,288.372000,0.529230 which respectively mean frame 2D BBOXz(x1,y1,x2,y2) score.

[russellyq]

@wangxiyang2022

Thanks for your reply.

As I read through the code. I found that calib_file is not used: https://github.com/wangxiyang2022/DeepFusionMOT/blob/a09ad79e1c0235c7a97d9c6c716c21efd94e5a65/tracking/matching.py#L157

[853108389]

I have the same question. Could you resolve this problem?

[BrickDanD]

For 3D detections 0,2,298.3125,165.1800,458.2292,293.4391,8.2981,1.9605,1.8137,4.7549,-4.5720,1.8435,13.5308,-2.1125,-1.7867 which respectively mean frame type(2 means car) 2D BBOXz(x1,y1,x2,y2) score 3D BBOX(h,w,l,x,y,z,rot_y) alpha.

For 2D detections 0,296.021000,160.173000,452.297000,288.372000,0.529230 which respectively mean frame 2D BBOXz(x1,y1,x2,y2) score.

So, what's the meaning of score in 3D detection results? If it means confidence level, Why is its value greater than 1 ？ Could you please answer this question? Thanks

[vandana1302238]

Hi @BrickDanD , did you get clarity on this. Can someone or @wangxiyang2022 help with this? Thanks!

[vandana1302238]

For 3D detections 0,2,298.3125,165.1800,458.2292,293.4391,8.2981,1.9605,1.8137,4.7549,-4.5720,1.8435,13.5308,-2.1125,-1.7867 which respectively mean frame type(2 means car) 2D BBOXz(x1,y1,x2,y2) score 3D BBOX(h,w,l,x,y,z,rot_y) alpha. For 2D detections 0,296.021000,160.173000,452.297000,288.372000,0.529230 which respectively mean frame 2D BBOXz(x1,y1,x2,y2) score.

So, what's the meaning of score in 3D detection results? If it means confidence level, Why is its value greater than 1 ？ Could you please answer this question? Thanks

Hi @BrickDanD , did you get clarity on this. Can someone or @wangxiyang2022 help with this? Thanks!