/cvpr2016

Learning Deep Representations of Fine-grained Visual Descriptions

Primary LanguageLuaMIT LicenseMIT

###Learning Deep Representations of Fine-grained Visual Descriptions Scott Reed, Zeynep Akata, Honglak Lee, Bernt Schiele

#####How to train a char-CNN-RNN model:

  1. Download the birds and flowers data.
  2. Modify the training script (e.g. train_cub_hybrid.sh for birds) to point to your data directory.
  3. Run the training script: ./train_cub_hybrid.sh

#####How to evaluate:

  1. Train a model (see above).
  2. Modify the eval bash script (e.g. eval_cub_cls.sh for birds) to point to your saved checkpoint.
  3. Run the eval script: ./eval_cub_cls.sh

#####Pretrained models:

#####Citation

If you find this work useful, please cite as follows:

@inproceedings{reed2016learning, 	
 title = {Learning Deep Representations of Fine-Grained Visual Descriptions,
 booktitle = {IEEE Computer Vision and Pattern Recognition},
 year = {2016},
 author = {Scott Reed and Zeynep Akata and Bernt Schiele and Honglak Lee},
}