Important instructions:

1) The supplementary materials contain two repositories:
	a) mmdetection3d - DensePointPillars and DenseCenterPoint implementation
	b) PillarNet - DensePillarNet implementation
	
2) Setup the conda environment using the steps mentioned in the README.md file of the respective repositories.

3) For the evaluation runs, check the instructions in the README.md file of the respective repositories.

Trained models:

Please download the checkpoint file from this URL:
https://drive.google.com/drive/u/5/folders/1QrsGYN3mKxB9W8yBmP7kPP7doCLSRPay

The checkpoint folder contains 5 checkpoint files:
	a) DensePointPillars with layerwise concatenation
	b) DensePointPillars with one shot aggregation (proposed method)
	c) DenseCenterPoint with layerwise concatenation
	d) DenseCenterPoint with one shot aggregation (proposed method)
	e) DensePillarNet with one shot aggregation (proposed method)
	
Log files:

The logs folder contains :
	a) Training logs of each of these methods
	b) Benchmark results (FPS)
	c) Model complexity analysis (FLOPS and model parameters)

State of the art results:
	a) DensePillarNet - https://evalai.s3.amazonaws.com/media/submission_files/submission_506914/c8621c15-1bc6-4a39-bd78-f029ec1b37eb.txt 
	b) DensePointPillars - Check screenshot in the supplementary materials
	c) DenseCenterPoint - https://evalai.s3.amazonaws.com/media/submission_files/submission_504521/f1cc8a85-ee6e-4c2d-936c-51813d002b44.txt