With the rapid development of emerging enterprises, efficient information resource sharing has become crucial. Entrepreneurial information contains various types of data, usually presented in images and visual documents, with rich spatial and temporal features. However, traditional information sharing models face difficulties in handling unstructured data and heterogeneous data from multiple sources, which limits the depth and breadth of information sharing. To address this issue, this study proposes an enterprise information resource sharing model based on YOLO algorithm, which combines a gated recurrent unit with an improved YOLOv4, taking shipyard enterprises as an example for research. YOLO, as a powerful real-time object detection tool, can effectively process image and visual document data. The results indicated that the proposed improved YOLOv4 achieved optimal recall, precision, and accuracy at a batch size of 16 and 80. It performed better than CNN and RNN in key metrics. MSE decreased by 25.33% and 5.17%, respectively, RMSE decreased by 13.64% and 2.58%, respectively, and MAE decreased by 56.67% and 26.37%, respectively. The YOLO-based model significantly enhances the efficiency and quality of information sharing in shipyard enterprises. Its applicability extends to other industries, providing scalable solutions for optimizing resource management.