Pose Estimation in 7 minutes - 30 FPS on CPU Tutorial

[@emmanueloluga9770]

This was very great, thank you for sharing

[@Augmented_AI]

I'm glad that I can help 😁. What else would you like to learn?

[@emmanueloluga9770]

@@Augmented_AI Awesome. What's the state of the Art and Gold Standard of Pose Estimation today?

[@nikhilchhabra]

Questions: Can we access and modify the architecture ? Can we retrain on new data to make it work on new edge cases ?

[@Augmented_AI]

Great question, I have not looked into this, but have a look at what is possible with the deeper API.

[@priyambordoloi771]

I am super ready about it. Also can you please specify whether it can detect multiple person's pose?

[@Augmented_AI]

So this one is just for a single person unfortunately. But they have ensured that it's quite accurate for single person pose estimation.

[@priyambordoloi771]

@@Augmented_AI yep I have applied the mediapipe, it works well for single person with fps around 22 .

[@benjaminselig1260]

Hello! There is no option to sign up on the Augmented StartUps website, how can I sign up?

[@denislambert8316]

Hello! I am wondering whether it is possible to export the value of each points of the pose from a video to analyze motions?

[@dred4947]

Brother can you make a vedio on. Pose estimation on a mp4vedio

[@moonlightericchien2378]

The performance is amazing. Can this model run on mobile (android system)?

[@Augmented_AI]

Yep it can😁

[@prathameshdeshmukh5954]

I want to print the detected pose..how can i? PLease help !

[@Scottdahockeyman15]

Hi I am new to this type of material, so please excuse my lack of knowledge. I am currently completing a project that requires me to install an API and use that API on a public dataset such as COCO to achieve similar results to a published paper. Is this possible using a CPU version such as in this video? or would the API have to be installed from source while using CUDA, for example?

[@dglalperen]

Is it possible to use this with a hololens 2 ?

[@youtubeallvideos5670]

Hello how do you transfer the estimation in to 3d model or rif

[@Nostalgia-futuro]

I have an instance segmentation model with MOT/ MOTS, I want to implement, a car pose estimation and a human pose estimation, how do we train the pose estimation, is there an annotation tool.

[@pyollowerjacvva]

Any luckk?

[@computervisionwithandy8181]

Can it run on embedded edge devices like Nvidia jetson nano? I am really waiting for it

[@Augmented_AI]

Yep it is possible because it is a light weight model

[@ll__LOVE__ll]

how to extract the animation to BLENDER??

[@olvaddeepfake]

can you export the animation?

[@androidpc9656]

sir why i m getting only 10 fps ,while detecting objects 6 fps, why this occurs ? i ve copied your code but no changes,

[@Augmented_AI]

What CPU do you have?

[@Augmented_AI]

Also what resolution are you using?

[@androidpc9656]

Processor : AMD A6 7310 APU with AMD Radeon R4 Graphics clocked at 2.0 Ghz. 4 core, x64 architecture Ram: 4gb resolution : 640x480

[@Augmented_AI]

It could be the performance of your CPU. Maybe try on a more powerful CPU and see if there is a change. I'm using an i7 coffeelake 3.8 Ghz CPU

[@priyambordoloi771]

cool

[@Augmented_AI]

Thanks Priyam and thank you for becoming a member :).