search

kminoda / VIODE

VIODE: A Simulated Dataset to Address the Challenges of Visual-Inertial Odometry in Dynamic Environments
92 stars 9 forks source link

Problem of RGB in rgb_ids.txt #8

Closed wjf1997 closed 3 years ago

wjf1997 commented 3 years ago

rgb_ids.txt provides correspondence between IDs and RGB values in semantic image. There are some problems. The calculated RGB value cannot match the value in the file, so I cann't determine which object this pixel should correspond to. For example, in parking_high sequence, the last dynamic vehicle's calculated RGB is 108, 91, 207, which cannot be found in rgb_ids.txt. Actually, IDs of six vehicles are from 245-251. In rgb_ids.txt, none of six sets of data is capable of matching 108, 91, 207. I drew six sets of color and fail to find correspondence compared to semantic image by naked eye.

kminoda commented 3 years ago

Thank you for pointing it out! (and again, sorry for the late reply) Let me investigate the problem, and hopefully give you a response in a week.

wjf1997 commented 3 years ago

Thank you for pointing it out! (and again, sorry for the late reply) Let me investigate the problem, and hopefully give you a response in a week.

Your help was very much appreciated.

kminoda commented 3 years ago

I've found out the problem, and updated the file with a newer version. It should be fine now, but let me know if you figure out any further problems.

Again, thank you very much for pointing this out!

weiningwei commented 2 years ago

Hello, I also met the same problem. For example, in the 3_high sequence of city_day, the rgb values of keyboard pointer's pixel position are is 144, 234,243 ; 113,119,231; 98,202, 227 in the three following figures. But I can't match the id in the rgbd_ids.txt file.

1669327872 the rgb value is 144, 234,243 309276595 the rgb value is 113,119,231 532633000 the rgb value is 98,202, 227