Mask R-CNN Object Segmentation in Python

[@NicolaiAI]

Join My AI Career Program www.nicolai-nielsen.com/aicareer Enroll in My School and Technical Courses www.nicos-school.com

[@aniketgangotri8740]

Hi I've been following your tutorial for finding Depth through stereo vision , I have a question , how do find the baseline between the 2 cameras ? In your Code In GitHub you have assumed the value of B as 9 but im not sure how you obtained this information .

[@nagendranlakshmi992]

I had the same question as well

[@NicolaiAI]

Hi this video is not really about stereo vision. But to find the baseline u simply just measure the distance between the two lenses of ur cameras. If u have a stereo camera the baseline is already provided in the specifications

[@aniketgangotri8740]

@@NicolaiAI Thanks for the quick reply ! Actually I have only a single a camera to work with , I have a setup where I move the camera by the baseline distance and take photos to emulate multiple camera setup . I think measuring the distance between the lenses is kind off crude , I actually tabulated the errors and they were increasing exponentially as I increased the baseline . What I'm looking for is to obtain the baseline as the translation of Camera 2 wrt to Camera 1 , do you have any idea how I can go about this ?

[@NicolaiAI]

@@aniketgangotri8740 u can try to estimate the camera transformation for every single frame by tracking points in the frames. Ill recommend you to watch my videos about Visual odometry

[@aniketgangotri8740]

@@NicolaiAI Thanks I'll check them out but I understand that if we decompose the Essential Matrix obtained through Stereo Calibration of 2 cameras we can theoretically obtain the translation matrix too ? From the translation matrix we can estimate the baseline right ? Do you have any tutorial recommendations for this approach to the problem ? Instead of 2 cameras I'll perform the calibration by using photos obtained from 2 different poses of the same camera

[@gayathrivenkat6366]

How do i get this code?Thank you

[@waleedak6164]

Many thanks to your video How can i export all mask to .csv file? with (false or 0) not detected and (true or 1) detected?

[@gokhanevlek8531]

Man if you talk about how we can make it real time I would be perfect!!!

[@NicolaiAI]

will look into that!

[@reubenthomas1033]

Yes please do! also do consider using a camera like the OAK-D, and running the algo on the intel compute stick inside the oak-d with the Luxonis depthai python library! That would be dope

[@daboomchakalaka7898]

Hi! Came across your channel today, and after watching some of your tensorflow transfer learning videos with colab I have instantly subscribed! I want to become a channel member, and pay for the help with a project subscription. Before I do I just want to ask, when is the soonest you will be available to provide help? If possible I'd like to have some help as soon as possible. I am working on a project utilising transfer learning of the mobilenet architecture for an autonomous driving project to predict an angle and a speed! :)

[@NicolaiAI]

Hi thanks a lot for watching! The way it works is that I help people out on the discord. I'll start to provide help immediately and then we just chat here and there when it's needed

[@daboomchakalaka7898]

@@NicolaiAI Nice! Where can I find the link to the discord? I will become a channel member today!

[@NicolaiAI]

@@daboomchakalaka7898 it will be under the community tab as a post only for members

[@daboomchakalaka7898]

@@NicolaiAI Cool man, I will become a member now then. Thank you for this youtube channel, it's the only machine learning channel I've seen that interacts so much with it's subscribers. Rare to see. Keep up the good work.

[@NicolaiAI]

@@daboomchakalaka7898 thanks a lot! Really appreciate it man