models

Pre-trained model Preparation

Please follow these instructions for pre-trained model preparation for training.

1. Download the Deeplab-v2 model trained for semantic segmentation on SketchyScene dataset here.

2. Create a directory named SketchyScene_DeepLabv2 under this root and place the model inside.

3. (Opt) We have also provided the code for other alternatives for sketch image feature extraction, such as FCN-8s, SegNet and Deeplab-v3+. But we have not provided the pre-trained models since they are too large. If you are in need of them, please email us :)