BROWSE
Detail
Conference Paper
Image Annotation Using VQGAN and Backtranslation
Cited 0 time in 
Share 
Share
- Authors
- Dong-Hyuck Im, Yong-Seok Seo
- Issue Date
- 2023-10
- Citation
- International Conference on Information and Communication Technology Convergence (ICTC) 2023, pp.1696-1698
- Publisher
- IEEE
- Language
- English
- Type
- Conference Paper
- Abstract
- The paper presents an image annotation technique based on VQGAN and backtranslation. Using the VQGAN model, it learns a quantized codebook that expresses an image in block units, encodes the image using the codebook, and then trains a back-translation model which translate image to text using small amount of text/image pair data. Using the back-translation model, we can generate synthetic text data from image only dataset. Experimental result shows that we can generate synthetic text describing Korean face images and successfully train a text-to-image generation model.
- KSP Keywords
- Back-translation, Experimental Result, Face Image, Generation model, Image annotation, Image generation, Translation Model, text data