Fig. 1. Diagram of YOLOv3 feature extraction and recognition

Fig. 2. Inception module

Fig. 3. Inception-ResNet module

Fig. 4. Diagram of darknet-139

Fig. 5. Curve of total loss

Fig. 6. Loss curve of 10000-50000 iterations

Fig. 7. Partial intermediate convolution results. (a) Layer 1; (b) layer 2; (c) layer 3; (d) layer 4; (e) layer 5; (f) layer 6; (g) layer 7; (h) layer 8; (i) original graph

Fig. 8. Target recognition effects. (a)(b) Tank target recognition; (c)(d) long-range rocket gun recognition; (e)-(h) antiaircraft gun recognition; (i) warship recognition

Fig. 9. Comparison of model recognition effect. (a)(b) YOLOv3; (c)(d) proposed model