Browse categories
Explore
Fiverr Pro
English
$
USD
Looks Like This Service Is On Hold
Cem Gunes is unavailable until Dec 26, 2026
“Hey, I'm currently a little busy, but I'll be happy to work with you when I'm available again.”
I will train or test 3d semantic segmentation models
Cem Gunes
About this gig
Hello!
I am an architect and computational designer who is advanced in deep learning tasks. Particularly, I have experience in 3D semantic segmentation. I provide deep learning solutions using pre-trained models or I give consultancy.
Additionally, I would help you with your custom datasets and training tasks. This task will require a ready-to-use dataset provided by a client.
As an input, the process requires a point cloud file to be segmented in a generic format. The result will be provided in .txt or .ply format.
Basic service does not include any script or deep learning model delivery. It offers delivery of a segmented point cloud.
Get to know Cem Gunes
Cem Gunes
Unavailable
- FromTurkey
- Member sinceMar 2022
- Last delivery2 years
Languages
English, Turkish, French, Italian
Cem Gunes is an architect and computational designer who is advanced in parametric design and digital fabrication-oriented scripting. He has expertise in C# and Python libraries of Grasshopper. He provides the most accurate and efficient solutions for tasks in different domains.
FAQ
Should I provide dataset?
Yes and no. If you want to have a generic deep-learning result for outdoor and indoor scenes, an additional dataset is not required. But if you need custom labels for your objects in the scene they should be provided.
What is the output?
For simple packages, a segmented point cloud will be delivered. For advanced packages, a deep learning model and a script will be delivered in addition to the results.
What should I provide in the first step?
For a simple semantic segmentation task, you need to send the 3D model of the scene. It is preferred in point cloud type but mesh models can be used as well.
Is there a point limit in a scene?
Yes, It is preferred to keep it under 50.000 points for the sake of training and testing performance. For unique cases, custom solutions will be applied such as splitting.
