Fig. 1. Structure of experimental platform

Fig. 2. Relative relationship of visual system

Fig. 3. Cladding system. (a) Preset powder; (b) powder spraying

Fig. 4. Three types of cladding state. (a) Good cladding; (b) uneven cladding; (c) not fused

Fig. 5. Model architectures. (a) Traditional ResNet architecture; (b) dual-channel ResNet architecture

Fig. 6. Image transformations used for augmentation. (a) Original image; (b) rotation; (c) random aspect ratio; (d) flipping

Fig. 7. Defect imaging enhancement process. (a) Original image; (b) removing noise; (c) dictionary matrix in K-SVD algorithm; (d) final image

Fig. 8. Flowchart of K-SVD

Fig. 9. Curves of accuracy and loss. (a) Accuracy curve; (b) loss curve

Fig. 10. Feature maps of ResNet features from different layers

Fig. 11. Rendering of Guided Grad-CAM and its flow chart for good cladding images