Recently, the semantics of scene text has been proven to be essential in fine-grained image classification. However, the existing methods mainly exploit the literal meaning of scene text for fine-grained recognition, which might be irrelevant when it is not significantly related to objects/scenes. We propose an end-to-end trainable network that mines implicit contextual knowledge behind scene text image and enhance the semantics and correlation to fine-tune the image representation. Unlike the existing methods, our model integrates three modalities: visual feature extraction, text semantics extraction, and correlating background knowledge to fine-grained image classification. Specifically, we employ KnowBert to retrieve relevant knowledge for semantic representation and combine it with image features for fine-grained classification. Experiments on two benchmark datasets, Con-Text, and Drink Bottle, show that our method outperforms the state-of-the-art by 3.72\% mAP and 5.39\% mAP, respectively. To further validate the effectiveness of the proposed method, we create a new dataset on crowd activity recognition for the evaluation. The source code and new dataset of this work are available at https://github.com/lanfeng4659/KnowledgeMiningWithSceneText.
@article{arxiv.2203.14215,
title = {Knowledge Mining with Scene Text for Fine-Grained Recognition},
author = {Hao Wang and Junchao Liao and Tianheng Cheng and Zewen Gao and Hao Liu and Bo Ren and Xiang Bai and Wenyu Liu},
journal= {arXiv preprint arXiv:2203.14215},
year = {2022}
}
Comments
Accepted to CVPR 2022. The source code and new dataset of this work are available at https://github.com/lanfeng4659/KnowledgeMiningWithSceneText