Chainer implementation for realismCNN proposed in Learning a Discriminative Model for the Perception of Realism in Composite Images.
Download pretrained caffe model
- Download pretrained caffe model.
python load_caffe_model.pyto transform pretrained caffe model into Chainer model.
Predict image's realism
- Download dataset Realism Prediction Data.
python mat2list_human_eval.pyto obtain image list & ground truth.
python predict_realism.pyto obtain prediction results. AUC score will be printed out, prediction score for each image will be stored in plain text file.
Image Editing towards generating more realistic composited images
- Download dataset Color Adjustment Data.
python mat2list_image_editing.pyto obtain image list.
python image_editing.pyto obtain more realistic images. (cut_and_paste image, generated image) will be saved in the
resultfolder, and a plain file will be generated recording (cut_and_paste loss, generated loss) for each image.
python [SCRIPT_NAME].py -hfor more options.
- Download converted Chainer VGG-19 model from here.
- If you want to download Transient Attributes Dataset, please see the project website for more details