Embedding_GAN

Embedding Adversarial Learning for Vehicle Re-Identification

We provide the embedding network without the adversarial learning, which provide the baseline method in our work. the code can be download from https://github.com/yanbai1993/Embedding-Network.

Requirements

• pytorch
• Linux
• Python 2 or 3
• CPU or NVIDIA GPU + CUDA CuDNN

Getting Started

Installation

• Install PyTorch and dependencies from http://pytorch.org
• Install Torch vision from the source.

git clone https://github.com/pytorch/vision
cd vision
python setup.py install

• Install python libraries visdom and dominate.

pip install visdom
pip install dominate

• Clone this repo:

git clone https://github.com/yanbai1993/Embedding_GAN.git
cd Embedding GAN

visualization for hard negative

Based on Embedding GAN, you can generate hard negative samples. To better illustrate the hard negatives generation procedure, we provided the test code and several models (six models under different training stages). The models can be downloaded from "https://pan.baidu.com/s/1vkmccegB5epCa48C7pANbQ". You need to put the models into 'checkpoints' dir.

sh test.sh

The generated images are saved in results dir.

performance test

VehicleID dataset

We provide the test code and models for VehicleID dataset, the model can be download from https://pan.baidu.com/s/1-l6d_VuQ0oXsx6W7PT0xMA. You can test the mAP and CMC result.

sh map.sh
sh cmc.sh

VeRI-776 dataset

For VeRI-776 dataset, we provide an evaluate script, "baseline_evaluation_FACT_776.m", on VeRi following https://github.com/VehicleReId/VeRidataset. You can download our dist file from https://pan.baidu.com/s/1SI9QfviZllwMzqVQbkhh9A. The training code and models will be released later.