search

s-gupta / rcnn-depth

Learning Rich Features from RGB-D Images for Object Detection and Segmentation
BSD 2-Clause "Simplified" License
365 stars 150 forks source link

Doubt in the fixed value of 31000 to estimate the horizontal disparity #48

Open dhorka opened 4 years ago

dhorka commented 4 years ago

Hi, I was going through the save_HHA code and I realize that to calculate the horizontal disparity you are using a fixed value 31000. I was wondering where this value comes from. I checked but I did not found the explanation for that. First I was thinking that it was a trick to obtain a value between 0 and 255, however, I realized that the maximum possible value of the disparity was 310, because you are forcing to be 100 the minimum value of pc(:,:,3). Is there any reason for this value?

I would appreciate if you can help with this doubt. @s-gupta

Best Regards,

Erientes commented 4 years ago

Did you figure out where the value comes from in the end? I had the same question and would like to use it on a different dataset for which fixed ( and unmotivated) values cause some problems.