Repository files navigation

• This is a pytorch implementation of the paper,Unsupervised Holistic Image Generation from Key Local Patches. (ECCV 2018).
• Paper link:https://arxiv.org/abs/1703.10730 (warning: this is an old version. Latest version will be uploaded!)

• Python2 or 3
• Cuda device (NVIDIA GTX1080Ti was used to test)
• Pytorch
• Visdom (optional)
• Tensorflow & Tensorboard (optional)

Download dataset via visitingcelebA orCompCar.

For celebA dataset,

You can download usingdownload.py

> python download.py celebA

For compcar dataset,Download the entire compcar dataset and some pre-processing is required.

You should crop the car patches using the ground truth bounding boxes, resize them128*128 resolution, and save them in a single directory.

We already extracted key patches from celebA and compcar dataset and save the bounding box coordinates tocelebA_allbbs.mat andcompcar_allbbs.mat.

You can extract key patches and use your own key patches.

Run

python main.py --db_name=celebA --dataset_root=YOUR_DATA_ROOT --is_crop=True --image_size=108 --output_size=64 --model_structure=unet

The resolution of output image can be enlarged by--output_size=128 or--output_size=256 options.

Run

python main.py --db_name=compcar --dataset_root=YOUR_DATA_ROOT --is_crop=False --image_size=128 --output_size=128 --conv_dim=64  --batch_size=32 --model_structure=unet

Modify the optionsoutput_size,conv_dim, orbatch_size to prevent out-of-memory error.