forked from Tongji-MIC-Lab/KAGS
-
Notifications
You must be signed in to change notification settings - Fork 0
/
README.txt
22 lines (16 loc) · 853 Bytes
/
README.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Source codes of KAGS
Requirements:
- Python 3.6
- pytorch = 1.3.0
- Other python packages: nltk, pycocotools, pyyaml, easydict, datasets
Data preparation:
For bottom-up feature, you can refer the instructions in https://github.com/peteanderson80/bottom-up-attention.
For fc or conv features, we directly download from https://vist-arel.s3.amazonaws.com/resnet_features.zip.
For commonsense knowledge, we utilize the ConceptNet API (https://github.com/commonsense/conceptnet5/wiki/API) for each concept word.
Training:
python train.py
Testing:
python test.py
Note that our experiments are conducted on a single 32GB TESLA V100 GPU.
Tengpeng Li, Hanli Wang, Bin He, Chang Wen Chen, Knowledge-enriched Attention Network with Group-wise Semantic for Visual Storytelling,
IEEE Transactions on Pattern Analysis and Machine Intelligence, accepted, 2023.