yashkant/concat-vqa
Official code for the paper "Contrast and Classify: Training Robust VQA Models" published at ICCV, 2021
Python
Issues
- 3
Corresponding images
#10 opened by JurijsNazarovs - 2
Script to generate negatives
#9 opened by JurijsNazarovs - 4
Description of data in splits folder
#7 opened by JurijsNazarovs - 1
CS score in the paper, table 2
#8 opened by JurijsNazarovs - 0
"For grounding truth file to extract features ", but extract_features.py didn't have --imdb_gt_file <path_to_imdb_npy_file_generated_above>
#6 opened by alice-cool - 0
"For grounding truth file to extract features ", but extract_features.py didn't have --imdb_gt_file <path_to_imdb_npy_file_generated_above>
#5 opened by alice-cool - 0
- 1
Extracting features Install vqa-maskrcnn-benchmark(link) repository and download the model and config.
#3 opened by alice-cool - 2
File missing
#2 opened by thaolmk54 - 2
Can't download the files from dropbox
#1 opened by PandaPYH