Fig. 1. Schematic of blade damage photoelectric detection device based on pulse laser synchronization

Fig. 2. (a) Design block diagram and (b) physical image of sub-nanosecond high-frequency pulse laser

Fig. 3. Output waveform of sub-nanosecond high-frequency pulse laser radiator. (a) Oscilloscope; (b) photoelectric detector

Fig. 4. Flowchart of large-format high-resolution image acquisition

Fig. 5. Leaf target extraction and damage detection framework based on deep learning

Fig. 6. Network architecture diagram of the YOLOv5

Fig. 7. Large format image segmentation process

Fig. 8. Partial data on inspection of wind turbine blades in wind farms

Fig. 9. Wind turbine blade damage dataset

Fig. 10. Comparison of proposed algorithm detection results. (a) Original images; (b) original YOLOv5 algorithm; (c) proposed algorithm

Fig. 11. P-R curves for damage target detection between proposed algorithm and other target detection algorithms. (a) Skin; (b) pollution; (c) breakage; (d) corrosion