A comprehensive code repository for AI & Robotics.
Bound to a long term development. Below is the list of present contents:
-
Datatypes (Pydantic based):
- Camera Intrinsic & Extrinsic
- Image
- Pose (SE3 & SO3)
-
PyTorch-Lightning Templates:
- Trainer
- MLP
- Dataloader
-
Distances:
- SO3 & SE3 manifold distances
-
Pointcloud:
- Furthest point sampling
- Ball query
-
Robotics:
- Stable inverse jacobian
- Planar & geodesic trajectory generation
-
Transformers:
- Upto 3D-Rotatory position embeddings
- Temporal-Image patch embedder
- Fast multi-head pooling attention module (MHA / MHPA)
- For Vision: ViT, MViT-v2 (auto-adaptable)
- For Video : MViT-v2
-
Visual object generalization:
- Dense Object Nets (DON)
- KeypointNet