Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
yangzhouxin 51dd4a165f | 3 years ago | |
---|---|---|
.. | ||
META-INF | 3 years ago | |
__pycache__ | 3 years ago | |
data | 3 years ago | |
eval_results | 3 years ago | |
log | 3 years ago | |
misc | 3 years ago | |
models | 3 years ago | |
scripts | 3 years ago | |
vis | 3 years ago | |
.DS_Store | 3 years ago | |
ADVANCED.md | 3 years ago | |
HSPLAN.py | 3 years ago | |
ImageCaptionModel.py | 3 years ago | |
LICENSE | 3 years ago | |
data_test.py | 3 years ago | |
dataloader.py | 3 years ago | |
dataloaderraw.py | 3 years ago | |
eval.py | 3 years ago | |
eval_ensemble.py | 3 years ago | |
eval_utils.py | 3 years ago | |
opts.py | 3 years ago | |
readme.md | 3 years ago | |
test-best.sh | 3 years ago | |
test-last.sh | 3 years ago | |
test.py | 3 years ago | |
test_HSPLAN.py | 3 years ago | |
train-wo-refining.sh | 3 years ago | |
train.py | 3 years ago | |
train.sh | 3 years ago |
基于transformer的图文识别模型(Image Caption Model)
该模型实现了从图像特征中生成文本描述的功能,通过从原始图像中提取的图像特征,可以生成根据图像特征描述该图像的语句。
代码基于pytorch框架,在生成语句阶段,可以输出语法正确的连贯语句。
CIDEr-D(Consensus-based Image Description Evaluation), BLEU(Bilingual Evaluation Understudy), METEOR(Metric for Evaluation of Translation with Explicit ORdering), ROUGE-L(Recall-Oriented Understudy for Gisting Evaluation), SPICE(Semantic Propositional Image Caption Evaluation)
使用MSCOCO数据集提取的用bottom up attention的预训练特征(https://github.com/peteanderson80/bottom-up-attention)
代码运行的环境与依赖:
类别 | 名称 | 版本 |
---|---|---|
操作系统 | Ubuntu | 16.04 |
编程语言 | Python | 3.6 |
底层驱动 | CUDA | 9.0 |
深度学习框架 | Pytorch | 1.0 |
代码的输入与输出。如下所示:
名称 | 说明 |
---|---|
输入 | fc_feats[图像中提取的整体特征],att_feats[图像中提取的局部特征],att_masks[图像特征的mask] |
输出 | 对应于图像特征的图像caption以及运行的时间 |
在terminal下运行以下命令。
cd project_dir
python HSPLAN.py
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》