Cheng Cui
Senior Engineer at Baidu
Head of Technology for Baidu PaddleOCR and PaddleX. He is responsible for the development of vision and OCR models and has contributed to the development of more than 80 models in the PaddlePaddle PP series, including PP-LCNet, PP-OCR, PP-YOLO, and RT-DETR. He has also participated in multiple computer vision projects within the company and holds over 30 domestic and international patents. He has won more than 10 gold medals or championships in international AI competitions, including several CVPR and ICCV workshop contests, and has been invited to give talks. In 2024, his project“Endangered Species AI Guardian 2.0”* won the **2025 Edison Award for Best New Product (Silver Award).
Topic
Latest Technologies and Industrial Applications of PaddleOCR
This talk will focus on OCR-related topics, introducing the new features of PaddleOCR 3.0 and how these features can be combined with large models for practical applications. It will first cover the current state of OCR and the challenges faced. Then, in response to these challenges, it will present the new features of PaddleOCR 3.0 (versions 3.0–3.3), including the next-generation general text recognition model PP-OCRv5, the next-generation document parsing tool PP-StructureV3, and the next-generation OCR+LLM key information extraction solution PP-ChatOCRv4. The talk will also explain how PaddleOCR’s MCP tools integrate with large models to become efficiency-boosting tools across industries. Finally, an industry application example of PaddleOCR combined with large models will be shared. Outline: Current state and challenges of OCR Introduction to PaddleOCR 3.0 Core technologies of PaddleOCR 3.0 Usage of PaddleOCR 3.0 and industrial applications
