Collect and discuss best algorithms for car detection and tracking

[mslavescu]

No description provided.

[angelocarbone]

See these kinds of articles, Mobileye specific:

https://scholar.google.ca/scholar?q=mobileye+vision+based+acc+paper

Also, check this new deep learning approach:

Geometric Deep Learning | Michael Bronstein || Radcliffe Institute
https://youtu.be/ptcBmEHDWds

Machine Learning Meets Geometry
https://youtu.be/pgomVDoEMrc

And then look at these kinds of techniques:

BoxCars: Improving Fine-Grained Recognition
of Vehicles using 3D Bounding Boxes
in Traffic Surveillance
https://arxiv.org/abs/1703.00686

SurfNet: Generating 3D shape surfaces using deep residual networks
https://arxiv.org/abs/1703.04079

This would be good for quick labeling of missed or wrongly labeled object:
https://github.com/scaelles/OSVOS-TensorFlow

I'll try to integrate this in Colaboratory:
https://github.com/CharlesShang/FastMaskRCNN