Commit 6b1f327b authored by tristan.kreuziger's avatar tristan.kreuziger
Browse files

Updated links

parent 834df37e
......@@ -24,14 +24,14 @@ If you use this code, please cite our paper given below:
[tqdm](https://tqdm.github.io/) and [NLTK](https://www.nltk.org/) packages should be installed. The code is tested with Python 3.7.4, PyTorch 1.4.0, torchvision 0.5.0, Pillow 6.1.0,
tqdm 4.36.1, NLTK 3.5 and Ubuntu 18.04.3.
* One of the [RSICD](https://github.com/201528014227051/RSICD_optimal), [Sydney-Captions](https://pan.baidu.com/s/1hujEmcG#list/path=%2F) and [UCM-Captions](https://pan.baidu.com/s/1mjPToHq#list/path=%2F) datasets should be downloaded.
* The pre-trained summarization model should be available. To train the pointer-generator model (defined [here](https://gitlab.tubit.tu-berlin.de/rsim/SD-RSIC/blob/master/summarization/models.py)) for summarization, you can use [this GitHub repository](https://github.com/rohithreddy024/Text-Summarizer-Pytorch).
* The pre-trained summarization model should be available. To train the pointer-generator model (defined [here](https://git.tu-berlin.de/rsim/SD-RSIC/-/blob/master/summarization/models.py)) for summarization, you can use [this GitHub repository](https://github.com/rohithreddy024/Text-Summarizer-Pytorch).
* Both the training and evaluation scripts except json files for the configuration arguments. Below, the arguments associated with training and evaluation are written, respectively.
## Training
* `exp_name`: Name of the experiment used while saving the model checkpoints.
* `images_dir`: Image directory of a dataset.
* `captions_file`: JSON file of a dataset that includes the captions of images.
* `vocabulary_path`: TXT file that includes vocabulary words. The vocabulary used during our experiments is provided in [vocab.txt](https://gitlab.tubit.tu-berlin.de/rsim/SD-RSIC/blob/master/vocab.txt) file.
* `vocabulary_path`: TXT file that includes vocabulary words. The vocabulary used during our experiments is provided in [vocab.txt](https://git.tu-berlin.de/rsim/SD-RSIC/-/blob/master/vocab.txt) file.
* `vocabulary_size`: Number of words used in the vocaulary.
* `emb_dim`: Image embedding size.
* `decoder_dim`: Decoder dimension for the first step of the SD-RSIC (Generation of Standard Captions).
......@@ -48,7 +48,7 @@ tqdm 4.36.1, NLTK 3.5 and Ubuntu 18.04.3.
* `exp_name`: Name of the experiment used while loading the model checkpoint.
* `images_dir`: Image directory of a dataset.
* `captions_file`: JSON file of a dataset that includes the captions of images.
* `vocabulary_path`: TXT file that includes vocabulary words. The vocabulary used during our experiments is provided in [vocab.txt](https://gitlab.tubit.tu-berlin.de/rsim/SD-RSIC/blob/master/vocab.txt) file.
* `vocabulary_path`: TXT file that includes vocabulary words. The vocabulary used during our experiments is provided in [vocab.txt](https://git.tu-berlin.de/rsim/SD-RSIC/-/blob/master/vocab.txt) file.
* `vocabulary_size`: Number of words used in the vocaulary.
* `emb_dim`: Image embedding size.
* `decoder_dim`: Decoder dimension for the first step of the SD-RSIC (Generation of Standard Captions).
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment