基于多任务联合预测的施工场景散料体积估算方法及装置
By employing a multi-task joint prediction method at the construction site, and utilizing a monocular RGB camera and a lightweight image semantic segmentation network for bulk material volume estimation, the problems of high computational resource consumption and insufficient measurement stability in existing technologies are solved. This achieves high-precision, lightweight bulk material volume estimation, which is suitable for resource-constrained equipment at construction sites.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIAMEN UNIV OF TECH
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, existing image processing methods cannot effectively solve the problem of bulk material volume estimation in construction scenarios. Furthermore, existing methods for estimating bulk material volume in construction sites are susceptible to environmental shadows. Existing laser scanning and multi-view vision solutions are easily affected by dust, vibration, and complex lighting conditions, and have high deployment costs. Monocular vision solutions consume significant computational resources, and measurement results are easily affected by environmental shadows. Moreover, the surface textures and natural accumulation characteristics of different bulk materials vary significantly, resulting in insufficient measurement stability.
A multi-task joint prediction method is adopted. The bulk material images acquired by a monocular RGB camera are preprocessed, and a lightweight image semantic segmentation network is used to output a binary mask to construct a structured feature vector representing the distribution of bulk materials. The volume, loading rate, and bulk material category probability are predicted by a multi-task deep learning model, and the final correction is performed by combining the effective volume of the storage and transportation container.
It achieves high-precision, lightweight bulk material volume estimation at construction sites, can adapt to dusty and complex lighting environments, reduces computing resource consumption, improves measurement stability and response speed, and is suitable for resource-constrained edge computing devices.
Smart Images
Figure CN122115541B_ABST