/PIXOR

Bird's Eye View Object Detection Algorithm for self-driving Cars

Primary LanguagePythonApache License 2.0Apache-2.0

PIXOR

Bird's Eye View Object Detection Algorithm for self-driving Cars

This is an ongoing attempt implement PIXOR with PyTorch for Univerity of Toronto's self-driving team (aUToronto)

The repository now contains the code of the model archeitecuture, the loss function, the training script, visualizaiton scripts and dataset wrapper.

The following sample detection are the result of a network trained on a Nvidia GTX-1080Ti for 10 epochs with a batch size of 4.

Image Sample

Ground Truth Ground Truth

Predictions Prediction

Output of the final layer Output

Credits

Project Contributors

  • Philip Huang
  • Allan Liu

Paper Citation below



@inproceedings{yang2018pixor,
  title={PIXOR: Real-Time 3D Object Detection From Point Clouds},
  author={Yang, Bin and Luo, Wenjie and Urtasun, Raquel}
}