Figures (15)  Tables (2)
    • Figure 1. 

      Relationship between labeling parameters and image.

    • Figure 2. 

      Precision of fire and smoke detection in the six algorithms.

    • Figure 3. 

      Recall rate of fire and smoke detection in the six algorithms.

    • Figure 4. 

      Precision-Recall curves of six fire detection algorithms.

    • Figure 5. 

      mAP of the six fire detection algorithms.

    • Figure 6. 

      F1 curves of six fire detection algorithms.

    • Figure 7. 

      F1 of six fire detection algorithms.

    • Figure 8. 

      Detection effect of the six algorithms for a building fire.

    • Figure 9. 

      Detection effect of the six algorithms for a single overturned flame.

    • Figure 10. 

      Detection effect of the six algorithms for multiple fire and smoke targets.

    • Figure 11. 

      Detection effect of YOLOv5 + TR + αCIOU algorithm in Bowfire dataset.

    • Figure 12. 

      Fire detection effects of the YOLOv5 + TR + αCIOU algorithm for different types of scenarios.

    • Figure 13. 

      Incipient stage of the factory fire.

    • Figure 14. 

      Developing stage of the factory fire.

    • Figure 15. 

      Fully developed stage of the factory fire.

    • True value
      Positive
      (real target)
      Negative
      ( non-target)
      Predicted valuePositiveTrue Positive (TP)False Positive (FP)
      NegativeFalse Negative (FN)True Negative (TN)

      Table 1. 

      Confusion matrix.

    • ModelClassPRMAP
      @0.5
      F1FPS/Frame
      per second
      Weight/
      MB
      YOlOv5All0.7780.5400.6410.6464.114.5
      Fire0.8590.6550.764
      Smoke0.6960.4260.518
      YOlOv5
      + CAC3
      All0.7760.5850.6530.6672.513.8
      Fire0.8290.6880.774
      Smoke0.7220.4810.531
      YOlOv5
      + TRC3
      All0.8550.5810.6970.6954.614.5
      Fire0.9030.6990.797
      Smoke0.8060.4630.597
      YOlOv5
      + αCIOU
      All0.7740.5830.6510.6661.314.5
      Fire0.8180.6670.765
      Smoke0.7290.5000.538
      YOlOv5
      + CA
      + αCIOU
      All0.7270.6140.6730.6760.613.8
      Fire0.8320.7100.794
      Smoke0.6220.5190.553
      YOlOv5
      + TR
      + αCIOU
      All0.8390.6850.7240.7058.814.5
      Fire0.8600.7100.806
      Smoke0.8180.5000.641

      Table 2. 

      Experimental results based on YOLOv5 and 5 optimization algorithms.