Pytorch implementation of Tips and Tricks for Visual Question Answering: Learnings from the 2017 Challenge by Teney et al.

• Python 2.7+
• NumPy
• PyTorch
• tqdm (visualizing preprocessing progress only)
• nltk (and this to tokenize questions)

• VQA 2.0
• COCO 36 features pretrained resnet model
• GloVe pretrained Wikipedia+Gigaword word embedding

• For questions and answers, go to data/ folder and execute preproc.py directly.
• You'll need to install the Stanford Tokenizer, follow the instructions in their page.
• The tokenizing step may take up to 36 hrs to process the training questions (I have a Xeon E5 CPU already), write a pure java code to tokenize them should be a lot faster. (Since python nltk will call the java binding, and python is slow)
• For image feature, slightly modify this code to convert tsv to a npy file coco_features.npy that contains a list of dictionaries with key being image id and value being the feature (shape: 36, 2048).
• Download and extract GloVe to data/ folder as well.
• Now we should be able to train, reassure that the data/ folder should now contain at least:

  - glove.6B.300d.txt
  - vqa_train_final.json
  - coco_features.npy
  - train_q_dict.p
  - train_a_dict.p
  

• (Update) For convenience, here is the link to tokenized questions vqa_train_toked.json and vqa_val_toked.json, make sure you run data/preproc.py to generate vqa_train_final.json, train_q_dict.p, etc.

Use default parameters:

python main.py --train

Train from a previous checkpoint:

python main.py --train --modelpath=/path/to/saved.pth.tar

Check out tunable parameters:

python main.py

python main.py --eval

This will generate result.json (validation set only), format is referred to vqa evaluation format.

• The default classifier is softmax classifier, sigmoid multi-label classifier is also implemented but I can't train based on that.
• Training for 50 epochs reach around 64.42% training accuracy.
• For the output classifier, I did not use the pretrained weight since it's hard to retrieve so I followed eq. 5 in the paper.
• To prepare validation data you need to uncomment some line of code in data/preproc.py.
• coco_features.npy is a really fat file (34GB including train+val image features), you can split it and modify the data loading mechanisms in loader.py.
• This code is tested with train = train and eval = val, no test data included.
• Issues are welcome!

• The paper.
• Their CVPR Workshop slides.