I was inspired by the Custom Segmentation template guide.

What this lens does is detect cups and then apply a high heat effect.

I built this project following steps shown below:

1) I followed all the instructions in the Custom Segmentation template guide; 2) then open the segmentation_training.ipynb notebook; 3) I opened google colaboratory; 4) I made some small changes to segment cups and you can see in this link segmentation_training.ipynb; 5) Following the steps in step 1, I downloaded the "melt_cup.onnx" model, then I opened the Lens Studio program and loaded this model, which you can download at this link melt_cup; 6) Finally I have added the heat distortion effect to my project, 7) Lens Studio project you can download at this link Melt Cup - Github; and 8) My Lens Project is public, unlocked and you can get at this link Melt Cup Lens

Challenges we ran into:

Honestly, they were complicated challenges, but satisfying to do! 1) Google Colaboratory offers little time to experiment for free (4-5 hours per day); 2) Lens Studio I did not find enough technical support to solve technical problems, I hope they publish some tutorial or more documentation on machine learning (how to solve technical problems, how to make a machine learning template, how to use other machine learning tools, for example kaggle); and 3) there are no machine learning webinars with Lens studio and taught by technical staff.

Accomplishments that we're proud of:

I finally learned to use Machine Learning to make Lens, now I just have to experiment with more templates or publish more templates or tutorials or webinars.

What we learned?

I've learned the basics of Lens Studio, and hope to move up to expert level.

What's next for Melt Cup?

It remains for me to experiment with other effects such as melting or deforming or crashing a cup. I don't know, but I would like technical staff to do a webinar to teach me how to make effects and load them into Lens Studio.

Built With

  • google-colaboratory
  • lens-studio
  • python
Share this project:

Updates