Face image segmentation for bicolor 3D printing

hdlbq · February 18, 2025, 5:17pm

Hi,
I’m new on this forum. Thanks for your patience and your help.
I have owned a Jetson Nano for a while, and think to use it for a problem I tried to solve for a long time.
I tried to print images (like faces) on a 3D printer, using only two colors (for example black and white), starting from color pictures.
I succeed in some cases (like in the picture joined), but at a huge cost: the time I spent.
I started by converting the picture in black and white,
Then I convert it to have segments (SVG format for example)
Then I import this picture in my favorite 3D modeling software (fusion360)
Finally, I select manually the huge number of small zones that should be printed in black, leaving the other zones in white.
My idea is to use the Jetson Nano to do that automatically (or with the minimal human help).
I searched with Google, but failed to find any similar project.
I will be grateful for any pointer to valuable resources that can help me

AastaLLL · February 19, 2025, 2:12am

Hi,

You can find some segmentation example in the below GitHub:

github.com/dusty-nv/jetson-inference

docs/segnet-console-2.md

master

<img src="https://github.com/dusty-nv/jetson-inference/raw/master/docs/images/deep-vision-header.jpg" width="100%">
<p align="right"><sup><a href="detectnet-tracking.md">Back</a> | <a href="segnet-camera-2.md">Next</a> | </sup><a href="../README.md#hello-ai-world"><sup>Contents</sup></a>
<br/>
<sup>Semantic Segmentation</sup></s></p>

# Semantic Segmentation with SegNet
The next deep learning capability we'll cover in this tutorial is **semantic segmentation**.  Semantic segmentation is based on image recognition, except the classifications occur at the pixel level as opposed to the entire image.  This is accomplished by *convolutionalizing* a pre-trained image recognition backbone, which transforms the model into a [Fully Convolutional Network (FCN)](https://arxiv.org/abs/1605.06211) capable of per-pixel labeling.  Especially useful for environmental perception, segmentation yields dense per-pixel classifications of many different potential objects per scene, including scene foregrounds and backgrounds.

<img src="https://github.com/dusty-nv/jetson-inference/raw/master/docs/images/segmentation.jpg">

[`segNet`](../c/segNet.h) accepts as input the 2D image, and outputs a second image with the per-pixel classification mask overlay.  Each pixel of the mask corresponds to the class of object that was classified.  [`segNet`](../c/segNet.h) is available to use from [Python](https://rawgit.com/dusty-nv/jetson-inference/master/docs/html/python/jetson.inference.html#segNet) and [C++](../c/segNet.h).  

As examples of using the `segNet` class, we provide sample programs C++ and Python:

- [`segnet.cpp`](../examples/segnet/segnet.cpp) (C++) 
- [`segnet.py`](../python/examples/segnet.py) (Python) 

These samples are able to segment images, videos, and camera feeds.  For more info about the various types of input/output streams supported, see the [Camera Streaming and Multimedia](aux-streaming.md) page.

See [below](#pretrained-segmentation-models-available) for various pre-trained segmentation models available that use the FCN-ResNet18 network with realtime performance on Jetson.  Models are provided for a variety of environments and subject matter, including urban cities, off-road trails, and indoor office spaces and homes.

This file has been truncated. show original

It demonstrates the multi-class use case.
For single class (face or non-face), the mask should be render in the black and white.

Thanks.

hdlbq · February 19, 2025, 12:48pm

Hi,
Thanks for your answer. And for the references.
I also think that a solution can be found using image segmentation.
Actually, I have currently two coasters: one success (my dog), that I made as a first try, and a failure (the head of my son). I spent a lot of time trying multiple software and many settings of them to find a good black and white image of the dog. That’s why I stopped for the 2nd coaster, trying to find a reliable way for doing such things.
And since I have a Jetson nano that sleeps on a shelf, it can be a good candidate.
I was a lecturer in computer science for 30 years, and I have still good skills in programming in many languages.
Best regards

AastaLLL · February 20, 2025, 1:59am

Hi,

In the above sample, the color is set in the below function:

github.com/dusty-nv/jetson-inference

c/segNet.cpp

master


      
          	net->loadClassColors(colors_path);

          

          	// set overlay alpha value

          	net->SetOverlayAlpha(SEGNET_DEFAULT_ALPHA);

          	

          	return net;

          }

          

          

          // loadClassColors

          bool segNet::loadClassColors( const char* filename )

          {

          	const uint32_t numClasses = GetNumClasses();

          	

          	if( !LoadClassColors(filename, &mClassColors, numClasses) )

          		return false;

          	

          	// allocate alpha flag mask

          	mColorsAlphaSet = (bool*)malloc(numClasses * sizeof(bool));

          

          	if( !mColorsAlphaSet )

So you can change it to black/white as output.
Thanks.

system · March 12, 2025, 3:03am

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.

Topic		Replies	Views
Segnet: Help to prepare image for better recognition Jetson Nano jetson-inference	6	1576	March 2, 2022
Will there be a centralised workflow for creating custom models for semantic segmentation on Jetson devices? Jetson Nano machine-learning , segmentation	4	215	May 21, 2024
Semantic Segmentation_Outdoor Navigation with Segnet and Jetson Nano Jetson TX2 jetson-inference	6	1014	October 18, 2021
Semantic Segmentation with SegNet Jetson Nano jetson-inference	5	1529	October 18, 2021
The output result of segnet-console.py is blurred. Jetson Nano	3	666	October 15, 2021
Dusty-nv sgmentation threshold setting option? Jetson Nano jetson-inference	6	383	June 14, 2023
Deeplab v3 semantic segmentation on TX2 Jetson TX2 jetson-inference	4	1409	October 18, 2021
Train Custom image Segmentation Model jetson_inference Jetson Nano ai-training	4	3182	October 15, 2021
Transfer Learning with PyTorch, I have a Problem Jetson Nano ai-training	10	823	April 26, 2021
Advice of semantic segmentation on agx Jetson AGX Xavier	6	1398	June 25, 2021

Face image segmentation for bicolor 3D printing

Related topics