fairness-captioning
fairness-captioning
0. Dataset preprocessing
Minimum settings to run show-attend-and-tell
Download the COCO dataset training and validation images. Put COCO dataset split JSON file from Deep Visual-Semantic Alignments in data/coco/. It should be named dataset.json
Run the preprocessing to create the needed JSON files:
python generate_json_data.py