Differentiable Product Quantization for Memory Efficient Camera Relocalization

Zakaria Laskar* · Iaroslav Melekhov* · Assia Benbihi · Shuzhe Wang · Juho Kannala

ECCV 2024

Paper | Project Page

We propose a hybrid scene compression method, D-PQED, which performs descriptor quantization-dequantization in an end-to-end differentiable manner. This approach is well-suited for structure-based localization methods, enabling accurate camera pose prediction under a very limited memory budget.

Citation

If you find our code or paper useful, please cite

@inproceedings{Laskar2024dpqed,
  author    = {Laskar, Zakaria and Melekhov, Iaroslav and Benbihi, Assia and Wang, Shuzhe and Kannala, Juho},
  title     = {Differentiable Product Quantization for Memory Efficient Camera Relocalization},
  journal   = {European Conference on Computer Vision (ECCV)},
  year      = {2024},
}

Acknowledgements

The Readme template was mainly inspired by the MonoSDF repository.