Load the cake/val_target. pkl

linjieli222 / VQA_ReGAT

Research Code for ICCV 2019 paper "Relation-aware Graph Attention Network for Visual Question Answering"

https://arxiv.org/abs/1903.12314

MIT License

180 stars 38 forks source link

Load the cake/val_target. pkl #25

Closed bobo123yhb closed 4 years ago

bobo123yhb commented 4 years ago

When the cake/val_target.pkl file is loaded, some labels appear with no content.But I looked at the Annotations in the official website, and saw that your answer differs from it.For example, question_id=393225000,answer=foodiebakercom.But your labels:[].There are also certain labels that appear differently than you gave them. For example, question_id:393225001,label=4 should correspond to 55.

linjieli222 commented 4 years ago

We follow ban-vqa to process the label2ans. To elaborate, those answers that appears < 9 times in the training set are discarded, which results in about ~3000 answers. Also, the answers are normalized as shown in tools/compute_softscore.py.

For more information, you can either directly visit the official repo of ban-vqa or read carefully about answer/label processing in Section 4.1 of our paper.

Thanks!