A data synthesizer for creating datasets of feet from a first-person perspective.
We threw this project together as part of the talk: Creating an Augmented Reality Football Game in a weekend. This project is currently setup to generate images and segmentation masks to be used in training a foot detector.
We have included a few feet, floors, and leg textures to get started training a prototype foot detector. This project should be taken as a jumping off point for data synthesis rather than a production-grade solution.
Open the xcode project, build, and run to start generating feet.
- FootRenderer/ (source for renderer)
- FootRenderer.xcodeproj (Xcode project)
- input/ (images used in synthesis: floors, feet)
- output/ (synthesized dataset)
This project was created by Laan Labs. We design and build solutions at the intersection of Computer Vision and Augmented Reality.
FootRenderer is released under the MIT License. See LICENSE for details.