Inspiration
3d printing is taking the world by storm! There are hundreds of companies, 3d printers are getting cheaper (Peachy printer is ~100$), but it is still cumbersome to make 3d models.
What it does
Take in a video of a scene and make 3d models of the objects in that scene.
How I built it
I built it by using ffmeg to segment a video into images, structure from motion and patched based multiview stero to do the scene reconstruction, and unsupervised machine learning to do object parsing.
Challenges I ran into
Getting a large system across multiple pieces of software to work.
Accomplishments that I'm proud of
The end to end system works reasonable well! I was able to get something awesome in 36 hours.
What I learned
I learned a lot about computer vision, data processing, and video formats.
What's next for 3d Object Parsing
Get it faster and easier to use. Get more accurate results.
Built With
- ffmeg
- meshlab
- osm-bundler
- python
Log in or sign up for Devpost to join the conversation.