The official pytorch implementation of FHENet:Lightweight Feature Hierarchical Exploration Network for Real-Time Rail SurfaceDefect Inspection in RGB-D Images.[PDF].The model structure is as follows：
Python 3.6, Pytorch 1.7.1, Cuda 10.2, TensorboardX 2.1, opencv-python.
If anthying goes wrong with environment, please check requirements.txt for details.
Baidu RGB-D 提取码: na4e
Table I Evaluation metrics obtained from compared methods. The best results are shown in bold.
| Models | Sm↑ | maxEm↑ | maxFm↑ | MAE↓ |
|---|---|---|---|---|
| DCMC | 0.484 | 0.595 | 0.498 | 0.287 |
| ACSD | 0.556 | 0.670 | 0.575 | 0.360 |
| DF | 0.564 | 0.713 | 0.636 | 0.241 |
| CDCP | 0.574 | 0.694 | 0.591 | 0.236 |
| DMRA | 0.736 | 0.834 | 0.783 | 0.141 |
| HAI | 0.718 | 0.829 | 0.803 | 0.171 |
| S2MA | 0.775 | 0.864 | 0.817 | 0.141 |
| CONET | 0.786 | 0.878 | 0.834 | 0.101 |
| EMI | 0.800 | 0.876 | 0.850 | 0.104 |
| CSEP | 0.814 | 0.899 | 0.866 | 0.085 |
| EDR | 0.811 | 0.893 | 0.850 | 0.082 |
| BBS | 0.828 | 0.909 | 0.867 | 0.074 |
| DAC | 0.824 | 0.911 | 0.875 | 0.071 |
| CLA | 0.835 | 0.920 | 0.878 | 0.069 |
| Ours | 0.836 | 0.926 | 0.881 | 0.064 |
Table II Test results of the performance of the relevant methods. The best results are shown in bold.
| Models | DCMC | ACSD | DF | CDCP | DMRA | HAI | S2MA | CONET | EMI | CSEP | EDR | BBS | DAC | CLA | Ours |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Pre↑ | 66.16% | 55.93% | 78.88% | 73.07% | 80.36% | 73.90% | 76.91% | 86.85% | 82.65% | 85.29% | 85.32% | 86.27% | 86.71% | 87.27% | 87.22% |
| Rec↑ | 25.46% | 63.88% | 31.02% | 36.14% | 74.18% | 91.67% | 82.83% | 78.61% | 87.76% | 87.61% | 86.60% | 87.31% | 88.09% | 86.59% | 88.34% |
| F1↑ | 33.36% | 55.65% | 42.12% | 44.98% | 74.84% | 78.98% | 78.20% | 80.55% | 83.31% | 85.14% | 84.12% | 85.63% | 86.23% | 86.07% | 87.01% |
| IOU↑ | 19.23% | 40.63% | 22.41% | 27.86% | 62.96% | 68.91% | 70.39% | 70.57% | 74.82% | 76.65% | 75.39% | 77.27% | 77.77% | 77.87% | 78.93% |
If you use FHENet in your academic work, please cite:
@article{zhou2023fhenet,
title={FHENet: Lightweight Feature Hierarchical Exploration Network for Real-Time Rail Surface Defect Inspection in RGB-D Images},
author={Zhou, Wujie and Hong, Jiankang},
journal={IEEE Transactions on Instrumentation and Measurement},
year={2023},
publisher={IEEE}
}
Model weights loading: Baidu 提取码：01xe