search

felixchenfy / Monocular-Visual-Odometry

A simple monocular visual odometry (part of vSLAM) by ORB keypoints with initialization, tracking, local map and bundle adjustment. (WARNING: Hi, I'm sorry that this project is tuned for course demo, not for real world applications !!!)
MIT License
390 stars 90 forks source link

What if we don't have the ground-truth to scale the translation vector ? #13

Open ChetanPatil28 opened 3 years ago

ChetanPatil28 commented 3 years ago

Hi @felixchenfy , this is such good work :-)

I have a doubt on 

Change scale:
Scale the translation t to be the same length as the ground truth so that I can make a comparison with ground truth. Then, scale the map points correspondingly

What if we don't have the ground-truth to scale the translation vector. I know that in such cases we can estimate the relative scale by using three-views, but I don't find a good resource on it. Can you point me out to any code that implements the relative-scale estimation ?

Thanks

felixchenfy commented 3 years ago

Hi, sorry I don't know what this is: "estimate the relative scale by using three-views" I was thinking to use Lidar to estimate the groundtruth. Don't know what "three-views" means.

On Sat, Nov 7, 2020 at 6:16 AM Chetan Patil notifications@github.com wrote:

Hi @felixchenfy https://github.com/felixchenfy , this is such good work :-)

I have a doubt on

Change scale: Scale the translation t to be the same length as the ground truth so that I can make a comparison with ground truth. Then, scale the map points correspondingly

What if we don't have the ground-truth to scale the translation vector. I know that in such cases we can estimate the relative scale by using three-views, but I don't find a good resource on it. Can you point me out to any code that implements the relative-scale estimation ?

Thanks

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/felixchenfy/Monocular-Visual-Odometry/issues/13, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFJY7WIXWQ64DHYUEJJIB4DSOVJFXANCNFSM4TNVTJKA .

mhyoosefian commented 3 years ago

Hi. Hope you are doing well. In this case, we have to use another sensor to overcome the scale ambiguity problem. For instance, it is common to fuse data from an IMU with data from the camera to estimate the scale. There are numerous resources on the internet, like: this one