Repository navigation
MobiLiteNet is a lightweight deep learning framework designed for real-time road distress detection on mobile and mixed reality (MR) devices. By integrating Efficient Channel Attention (ECA), structural refinement, sparse knowledge distillation, structured pruning, and quantization, the model achieves high detection accuracy with significantly reduced computational cost. Field tests confirm its applicability in mobile road monitoring, while a diverse dataset from Europe and Asia ensures robustness. Experimental results show that MobiLiteNet outperforms baseline models, enabling scalable and accurate road distress detection for intelligent transportation and infrastructure management.
Key findings include:
Proposed MobiLiteNet, a novel lightweight deep learning framework optimized for real-time road distress detection on mobile and MR devices.
Successful deployment on MR devices in road engineering projects in Aachen, Germany, demonstrating practical feasibility in complex environments.
Constructed a diverse dataset with road distress images from Europe and Asia, enhancing model robustness and generalization.