DATA contains the training/testing pairs and pose estimations of our method.

PG2 denotes the codes of paper Pose Guided Person Image Generation

Deform denotes the codes of paper Deformable GANs for Pose-based Human Image Generation

VUNet denotes the codes of paper A Variational U-Net for Conditional Appearance and Shape Generation

Samples contains several tested samples using the given codes and models.

• Download market dataset https://drive.google.com/file/d/0B8-rUzbwVRk0c054eEozWG9COHM/view. Unzip this file to a folder. Rename this folder to market-dataset. Rename bounding_box_test and bounding_box_train with test and train.

• Download deep fasion dataset in-shop clothes retrival benchmark. You will need to ask a password from dataset maintainers. Move img/ to data folder and rename it fashion/. Our key-point estimations are in DATA. Run script Deform/data/split_fasion_data.py to randomly split the data into training and testing sets.

• Data preparation for PG2 can be referred to the TF-record data preparation steps in PG2/README.md

• To test a individual method, please read the README.md file in the corresponding folder first.