Object Recognition via Geometric Shapes

Job ID: 36826707

Budget: $1,500 – $3,000 USD

Testing a New Approach to Object Recognition in Virtual Environment

1) Create a basic artificial environment; room or display platform that has specific dimensions and grid like coordinates throughout its height, length, and width and flexible
dimension settings
2) Import 3D Objects into this 3D Environment in a standard format and assign dimensions that reflect approximate real-life size & volume
3) Convert these 3D objects into an amalgamation of 3D geometric shapes; Sphere, Half Sphere, Cone, Cylinder, Triangular Prism, Cuboid, Square Pyramid, Tetrahedron, Truncated Cylinder and Torus.
Minimum number of geometric shapes to cover maximum amount of 3D object inner space/volume. Max 10% of object volume can remain empty
4) Identify, compare and classify 3D objects based on set of geometric shapes they contain. You can neural networks (function optimization) or rules I prepared; shape type, size, count, intersection angle and such
I posted this project a number of times before, but most applicants were unaware of project requirements and unwilling or unable to follow my rules so I am also including neural networks as an option as long as it works based on 3D geometric shapes or you can follow the rules in the attached file

Project Budget: $1,600
Related categories: Python Neural Networks Pytorch Computer Vision Pandas