junaidcs032 / motbeyondpixels Goto Github PK

Hi! Thanks for putting up your code. I wasn't able to find the filtering scripts for both rccnet and subcnn detections in the downloaded Data dir. Would it be possible to share them? Thank you!

I can't find the code about "Shape and Pose cost".
Does this repository include its code?

Thanks!

In main_script_train.m (main_script_test.m), the number "44" is used for scaling the motion.
https://github.com/JunaidCS032/MOTBeyondPixels/blob/master/src/main_script_train.m#L167
I think "1.72" is the height of camera, but I don't know about "44".

Thanks!

Did you evaluate the method using 'all training data' in Table 3 ?
If no, could you provide additional information?

Thanks in advance,

Yuki Tsuji

Thank you very much for sharing your code!

Question -- how do you create the 2D features that are saved in Features2D_mat? I may have missed the description of the feature extraction algorithm. Could you also share this code?

Hello. Firstly, thank you for providing the code.

Here is one question about the camera motion.
To compensate one frame of ego motion, I expected the scale of camera motion values should be meters per 0.1 sec (as KITTI is 10 frames/sec).
But, when I check the camera motion values from ORB slam and scaling (1.72/44), the values seem to be too small.
Could you explain why such a small value?

RRC_Detections_txt/train/15 is missing.
This scene contains cars so it shouldn't be empty.
Thank you.

Thank you very much for sharing your code!
Question --- how to get motion data?I want to use ur code apply in my data,but i don't know how to get it.Could you tell how to get it?

I am sorry to bother you. I want to know how you use the KITTI test set to evaluate your algorithm. I mainly want to know the label file of the KITTI test set because I did not find the label file of KITTI's test set.