this is a re-implementation of Recurrent Models of Visual Attention in tensorflow of lim0606/lasagne-ram, which can run on multi-gpus
first, go to the utils/reader/read_img/
dir, and run imgs_to_records.py --src src_dir --dst dst_dir
to make your own dataset to tfrecords
second, in the root dir, run python train.py
to train on your own dataset
#reference