Ask_Attend_and_Answer

Ask, Attend and Answer: Exploring Question-Guided Spatial Attention for Visual Question Answering

Code

Instructions for training the "SMem-VQA Two-Hop" model:

1. Download the provided caffe folder and install caffe following the instructions in http://caffe.berkeleyvision.org/installation.html .

2. Run ./example/data/get_image.sh to download MSCOCO image data.

3. Run ./train/train_mm.sh to train the model.

Citation

@article{xu2015ask, title={Ask, Attend and Answer: Exploring Question-Guided Spatial Attention for Visual Question Answering}, author={Xu, Huijuan and Saenko, Kate}, journal={arXiv preprint arXiv:1511.05234}, year={2015} }