Yang Feng

Path to https://github.com/tensorflow/models

链接: https://pan.baidu.com/s/1m-p7aTF0L_mFzPMdKK7btw 提取码: nzqq

I will not maintain the crawl_descriptions.py file. You need to change it according to the Shutterstock website.

You should read the code at https://github.com/fengyang0317/unsupervised_captioning/blob/master/config.py

> model_name = 'faster_rcnn_inception_resnet_v2_atrous_oid_2018_01_28' path_to_ckpt = (TF_MODELS_PATH + '/research/object_detection/' + model_name + '/frozen_inference_graph.pb'tensorflow.python.framework.errors_impl.NotFoundError: /home/fy/PycharmProjects/unsupervised_captioning-master/workspace/tf_models/research/object_detection/faster_rcnn_inception_resnet_v2_atrous_oid_2018_01_28/frozen_inference_graph.pb; No such file or directory how to solve the problem? Looking forward to your reply. >...

The following is a working conda environment. You create a yml file with the following content and then create a conda environment using the yml. ``` name: caption2 channels: -...

Replace master with archive https://github.com/tensorflow/models/tree/archive

The parameters are shown in the code or training commands. If you are training without initialization, you should train at least 20000 iterations. By the way, the checkpoints are released...

Are you training using the following command? ``` python im_caption_full.py --inc_ckpt ~/workspace/ckpt/inception_v4.ckpt\ --multi_gpu --batch_size 512 --save_checkpoint_steps 1000\ --gen_lr 0.001 --dis_lr 0.001 ``` In Google drive, saving folder contains the checkpoints...

I think the batch_size 256 is ok. You may need to reduce the learning rate too. Can you try 5e-4?