In computer vision, scene text detection and recognition are essential domains, particularly for real-time processing applications like autonomous cars and security surveillance. Despite Vietnamese being widely spoken, research and development of systems tailored for Vietnamese scene text remain limited. This paper explores deep learning-based approaches for detecting and recognizing Vietnamese text in scene text images. We evaluated state-of-the-art text detection models and found that the Character Region Awareness for Text Detection (CRAFT) model achieved the best performance, with 90% H-mean on the VinText dataset. For text recognition, we fine-tuned the VietOCR model on the VinText dataset, resulting in a character-level accuracy of 87% and a word-level accuracy of 79%. These findings demonstrate the potential applications of our framework in areas such as digital document management, assistance for visually impaired individuals, and automated translation systems.