Task Progressive Curriculum Learning: Approach for Robust Visual Question Answering

We provide the PyTorch implementation for Task Progressive Curriculum Learning: Approach for Robust Visual Question Answering (BMVC 2025).

Dependencies

• Python version: 3.12.0
• PyTorch version: 2.1.1
• dependencies in requirements.txt
• We train and evaluate all of the models based on one Nvidia H100 GPU using 48Gb of memory.

Getting Started

Installation

1. Clone this repository:

    git clone 
    cd TPCL
   

2. Install PyTorch and other dependencies:

    pip install -r requirements.txt
   

Download and preprocess the data

cd data 
bash download.sh
python preprocess_features.py --input_tsv_folder xxx.tsv --output_h5 xxx.h5
python feature_preprocess.py --input_h5 xxx.h5 --output_path trainval 
python create_dictionary.py --dataroot vqacp2/
python preprocess_text.py --dataroot vqacp2/ --version v2
cd ..

Training

• Train LXMERT with Dynamic TPCL

CUDA_VISIBLE_DEVICES=0 python src/main_tpcl_dyn.py --mode lxmert --cl linguistic --dataset cpv2 --output 'output/' --img_root /vqacp2/coco/ --dataroot /vqacp2/

• Train LXMERT with Fixed TPCL

CUDA_VISIBLE_DEVICES=0 python src/main_tpcl_fixed.py --mode lxmert --cl linguistic --dataset cpv2 --output 'output/' --img_root /vqacp2/coco/ --dataroot /vqacp2/

• Train the LXMERT with TPCL with 30% of the original training set

CUDA_VISIBLE_DEVICES=0 python src/main_tpcl_dyn.py --ratio 0.3 --mode lxmert --cl linguistic --dataset cpv2 --output 'output/' --img_root /vqacp2/coco/ --dataroot /vqacp2/ --ratio 0.3 

Evaluation

• A json file of results from the test set can be produced with:

CUDA_VISIBLE_DEVICES=0 python test.py --dataroot data/vqacp2/ --img_root data/coco/ --checkpoint_path saved_models_cp2/best_model.pth --output saved_models_cp2/result/

3. Running: refer to Download and preprocess the data, Training and Evaluation steps in Getting Started.

Reference

If you found this code is useful, please cite our paper:

@inproceedings{Akl_2025_BMVC,
author    = {Ahmed Akl and Abdelwahed Khamis and Zhe Wang and Ali Cheraghian and Sara Khalifa and Kewen Wang},
title     = {Task Progressive Curriculum Learning for Robust Visual Question Answering},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_294/paper.pdf}
}

Acknowledgements

This repository contains code modified from D-VQA, thank you very much!