# Using DIGITS to create a new segmentation model???? 

**URL:** <https://forums.developer.nvidia.com/t/using-digits-to-create-a-new-segmentation-model/84283>\
**Category:** Jetson Nano\
**Created:** [November 3, 2019, 11:36pm UTC](https://forums.developer.nvidia.com/t/using-digits-to-create-a-new-segmentation-model/84283 "2019-11-03T23:36:32Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![allenabigail500](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@allenabigail500](https://forums.developer.nvidia.com/u/allenabigail500)\
**Post date:** [November 3, 2019, 11:36pm UTC](https://forums.developer.nvidia.com/t/using-digits-to-create-a-new-segmentation-model/84283/1 "2019-11-03T23:36:32Z")

</div>

I bit off so much more than I can chew with this project, but I was hoping I could get some help here. I’ve been trying to use the Jetson Nano to be able to read and understand American Sign Language, but clearly I’m stuck. I am trying to see if there’s already a pre-trained segmentation model for hands specifically (I haven’t been able to find one) or how I would go about creating one. This is my first project with a Jetson Nano, and my first major coding endeavor with Python. I know I bit off more than I can chew, but is there anyone out there who knows how I could find/create a trained segmentation model for human hands specifically?

EDIT:  
I tried using DIGITS but can’t seem to be able to functionally use it, seeing as I can’t quite figure out how to use Docker to make a container. PLEASE HELP

---

<div class="post-metadata">

**Author:** ![AastaLLL](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aastalll/32/14043_2.png) [@AastaLLL](https://forums.developer.nvidia.com/u/AastaLLL)\
**Post date:** [November 6, 2019, 7:40am UTC](https://forums.developer.nvidia.com/t/using-digits-to-create-a-new-segmentation-model/84283/2 "2019-11-06T07:40:57Z")

</div>

Hi,

Sorry that we don’t have a segmentation model directly for HAND.  
But you can re-train a new one with the steps shared in following tutorial:  
[https://github.com/dusty-nv/jetson-inference/blob/master/docs/segnet-console-2.md](https://github.com/dusty-nv/jetson-inference/blob/master/docs/segnet-console-2.md)

In general, there are two steps required.

**1.** Train a segmentation model on a desktop GPU:  
It’s still recommended DIGITs for the training jobs for the good visualize function.  
Instead of docker, it’s also available to setup the DIGITs on your environment directly:  
[https://github.com/NVIDIA/DIGITS/blob/master/docs/BuildDigits.md](https://github.com/NVIDIA/DIGITS/blob/master/docs/BuildDigits.md)

**2.** Copy the model into Jetson and run it with _jetson\_inference_,

Thanks.

---

<div class="post-metadata">

**Author:** ![nadeemm](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/nadeemm/32/367020_2.png) [@nadeemm](https://forums.developer.nvidia.com/u/nadeemm)\
**Post date:** [October 15, 2021, 2:50am UTC](https://forums.developer.nvidia.com/t/using-digits-to-create-a-new-segmentation-model/84283/3 "2021-10-15T02:50:46Z")

</div>

Closed on October 15, 2021, 2:50am UTC


