Vehiclereid Baseline

This is a baseline framework for vehicle re-id on the VeRi dataset, implementated by PyTorch.

The backbone is ResNet-50 or ResNet-101 which is trained by cross entropy loss.

This is a very simple baseline for image-based vehicle re-id without any tricks, augmentation, metric learning methods, and use of license plates and spatio-temporal information. Therefore, the results may be very poor.

Dependencies

• PyTorch 1.0
• Python3