As a classic of Chinese traditional decorative art blending Buddhist culture, has been extensively utilized across various domains, the Baoxiang floral motif faces challenges in intelligent generation, style modernization, and design efficiency in modern applications. This study uses deep learning algorithms and semantic annotation to decode its cultural genes, employs LoRA lightweight training technology for controllable feature generation, and establishes a “traditional inheritance-modern application” evaluation framework with digital tools (EMD, ResNet-50, LPIPS, YOLOv5, etc.) and manual scoring to validate its effectiveness. Design practice shows human-AI collaboration makes the motif align with modern aesthetics, enhances designer efficiency, and boosts traditional culture preservation. The research offers a viable approach for traditional motif modernization and introduces a new paradigm for intelligent human-computer collaborative design and cultural heritage computing, facilitating the innovative transmission of intangible cultural heritage.