LaBERT
1.0.0
該存儲庫提供了論文長度可控圖像字幕的實現。
conda create --name labert python=3.7
conda activate labert
conda install pytorch=1.3.1 torchvision cudatoolkit=10.1 -c pytorch
pip install h5py tqdm transformers==2.1.1
pip install git+https://github.com/salaniz/pycocoevalcap
Bert-base
和 Faster-RCNN 的fc6
層。火車
python -m torch.distributed.launch
--nproc_per_node= $NUM_GPUS
--master_port=4396 train.py
save_dir $PATH_TO_TRAIN_OUTPUT
samples_per_gpu $NUM_SAMPLES_PER_GPU
繼續火車
python -m torch.distributed.launch
--nproc_per_node= $NUM_GPUS
--master_port=4396 train.py
save_dir $PATH_TO_TRAIN_OUTPUT
samples_per_gpu $NUM_SAMPLES_PER_GPU
model_path $PATH_TO_MODEL
推理
python inference.py
model_path $PATH_TO_MODEL
save_dir $PATH_TO_TEST_OUTPUT
samples_per_gpu $NUM_SAMPLES_PER_GPU
評價
python evaluate.py
--gt_caption data/id2captions_test.json
--pd_caption $PATH_TO_TEST_OUTPUT /caption_results.json
--save_dir $PATH_TO_TEST_OUTPUT
如果該專案對您的研究有幫助,請考慮在您的出版物中引用我們的論文。
@article{deng2020length,
title={Length-Controllable Image Captioning},
author={Deng, Chaorui and Ding, Ning and Tan, Mingkui and Wu, Qi},
journal={arXiv preprint arXiv:2007.09580},
year={2020}
}