Видео

Haha! That’s great!

Thank you!

Thank you for your kind words!

That's great to hear! Glad you found it helpful.

Thanks a lot!

Thank you! Cheers!

Thank you so much!

Yup, it’s not easy to understand these math equations. But I hope the video provides some intuition on why and how it works.

Thank you! Cheers!

Thanks! The animation comes from PowerPoints. I edit the video with Adobe premiere pro.

Glad it was helpful! Thanks a lot!

Glad you enjoyed it! Thanks for subscribing!

Probably not. LiDAR estimates the distance by detecting reflected light. It would have difficulty in shiny objects (since the specular surfaces reflect light at specific angle). So unless the sensor is perfectly aligned with the reflected path, it won’t “see” the object. Therefore, LiDAR would help in estimating the geometry of diffuse surface, but not shiny objects.

@jbhuang0604 good point mate, well I'd work in this field and see lidar uses polarization to specify the surface type and using dual wavelength emitters can help with transparent or translucent objects and at last there's a way called multi echoes processing which are to be done to distinguish between surface and it's reflectivity And using RGB data one can make a proper 3d reconstruction of the object and environment surrounding it... Just like apple does but not quite accurate.. I was in game development and i used these techniques to create 3d version of object but that time it was just limited to shape size and albedo but now it's far more developed..

Wow! That’s cool! Yup, LiDAR usually has difficulty in working with reflective and transparent surfaces. It’s interesting that you could handle these challenging cases.

That’s right!

Nope!

@jbhuang0604 just that you are presenting the algorithm as "we".

Ah! Got it. This is a collaborative work between Google and University of Maryland College Park.

Glad it was helpful!

@@jbhuang0604 Love all of the content on this channel! Thank you so much for doing this. I am spreading info about this great channel everywhere :))

Glad you liked it!

Glad you liked the video!

Very cool!

Thanks for the kind words, Xuan!

Thanks so much for your kind words!

What’s that?

@ ice coffee

I am a bit confused. The method is for general objects, not specific to people (or race).

@@jbhuang0604 yes as far as you know , hard to estimate what associations the network carries forward as you extrapolate capabilities over time , that may bake in bias in ways you had not considered

compounding interest of color schema as best practice could result in causal tendencies across varied modality as capacities increase toward more general purpose networks , for example -- were the masking purely done in black and white for autonomous vehicle , with black being labeled as a phantom person , an edge case scenario (1 in 1million) might result in a zoox or waymo determining its ok to smack me on the road

I am a huge fan of this work in any case , that's why I am here

Got it! Thanks for the reminder! Yes, we definitely need to work on reducing these unintentional bias.

Really appreciate your comment! Glad that you like the video!

Thanks for the feedback!

Thank you!

Thanks a lot!

Glad you like it! Thanks!

Thanks a lot for your comment. Your work is a big inspiration for us!

Thanks for your kind words!

Thank you! You made my day!

Thanks, I'm glad you enjoyed it!

Thank you!!

You’re welcome! Happy that you liked it.

Good catch! Clearly I was just trying to make sure if people are paying attention. :-p

We are working on it! Stay tuned!

Exciting times ahead!!

We are still working on a version that we can release.

We are working on it. Stay tuned!

We are working on it!

That’s COOL!

We are excited as well!

Thanks for your comment. We cannot release the version building upon Google internal pretrained model so we have to rebuild it using open source model. We are still working on it and hopefully can release that version.

@@jbhuang0604 Thats awesome, looking forward to try it! Take your time though; no need to rush. :) I can definitely wait.

Great! We are working on it!

Thank you so much!

Yes, we are very excited about what new opportunities this tool can unlock!

Thanks for the kind words!