SMILE: Sequence-to-Sequence Domain Adaption with Minimizing Latent Entropy for Text Image Recognition
This work addresses the challenge of recognizing text in real-world images without manual annotation, though it appears incremental as it builds on existing sequence-to-sequence models for domain adaptation.
The paper tackles the problem of domain shift between synthetic and real-world text images in text recognition by proposing an unsupervised domain adaptation method that minimizes latent entropy with class-balanced self-paced learning, achieving better recognition results than existing methods on most benchmarks.
Training recognition models with synthetic images have achieved remarkable results in text recognition. However, recognizing text from real-world images still faces challenges due to the domain shift between synthetic and real-world text images. One of the strategies to eliminate the domain difference without manual annotation is unsupervised domain adaptation (UDA). Due to the characteristic of sequential labeling tasks, most popular UDA methods cannot be directly applied to text recognition. To tackle this problem, we proposed a UDA method with minimizing latent entropy on sequence-to-sequence attention-based models with classbalanced self-paced learning. Our experiments show that our proposed framework achieves better recognition results than the existing methods on most UDA text recognition benchmarks. All codes are publicly available.