Awesome

Image Captioning with Deep Bidirectional LSTMs

This branch hosts the code for our paper accepted at ACMMM 2016 "Image Captioning with Deep Bidirectional LSTMs", to see Demonstration.

Features

• Training with Bidirectional LSTMs
• Implemented data augmentation: multi-crops, multi-scale, vectical mirroring
• Variant Bidirectional LSTMs: Bi-F-LSTM, Bi-S-LSTM

Usage and Example

• This work extends "Long-term Recurrent Convolutional Networks (LRCN)" to bidirectional LSTMs with data augmentation
• We provide an example flickr8K, in which you can train proposed networks
• (1) download flickr8 training and test images, and put it to "data/flickr8K/images/", the dataset splits can be found in "data/flickr8K/texts/"
• (2) create databases with "flickr8K_to_hdf5_data_forward.py" and "flickr8K_to_hdf5_data_backward.py"
• (3) train network with "multi_train_Bi_LSTM.sh"
• (4) perform image caption generation and image-sentence retrieval experiments with "bi_generation_retrieval.py"

Citation

Please cite in your publications if it helps your research:
@inproceedings{wang2016image,
title={Image captioning with deep bidirectional LSTMs},
author={Wang, Cheng and Yang, Haojin and Bartz, Christian and Meinel, Christoph},
booktitle={Proceedings of the 2016 ACM on Multimedia Conference},
pages={988--997},
year={2016},
organization={ACM}}

Following is orginal README of Caffe

Caffe

Caffe is a deep learning framework made with expression, speed, and modularity in mind. It is developed by the Berkeley Vision and Learning Center (BVLC) and community contributors.

Check out the project site for all the details like

• DIY Deep Learning for Vision with Caffe
• Tutorial Documentation
• BVLC reference models and the community model zoo
• Installation instructions

and step-by-step examples.

Please join the caffe-users group or gitter chat to ask questions and talk about methods and models. Framework development discussions and thorough bug reports are collected on Issues.

Happy brewing!

License and Citation

Caffe is released under the BSD 2-Clause license. The BVLC reference models are released for unrestricted use.

Please cite Caffe in your publications if it helps your research:

@article{jia2014caffe,
  Author = {Jia, Yangqing and Shelhamer, Evan and Donahue, Jeff and Karayev, Sergey and Long, Jonathan and Girshick, Ross and Guadarrama, Sergio and Darrell, Trevor},
  Journal = {arXiv preprint arXiv:1408.5093},
  Title = {Caffe: Convolutional Architecture for Fast Feature Embedding},
  Year = {2014}
}