Table 1: confusion matrix

TP: Positive samples predicted as positive by the model.

FP: Negative samples predicted as positive by the model.

FN: Positive samples predicted as negative by the model.

TN: Negative samples predicted as negative by the model.

Algorithm	MIT Indoor67	SUN397
MR-CNNs	85.40	77.64
MR-CNNs+ObjNet+Sum	86.21	78.10
MR-CNNs+ObjNet+Concat	86.73	78.26
The algorithm in this article	88.40	79.75

Table 2: The accuracy of this algorithm and other algorithms on the MIT Indoor67 and SUN397 datasets

Table 3: Comparison of the combined effect of attention and Context Gating

Table 1

Table 2

Table 3

Figure 1

Figure 2

Figure 3

Figure 4

Figure 1: Attention mechanism

Figure 2: Network Structure

Figure 3: Feature Conversion

Figure 4: Examples of some scenarios in the MIT Indoor67 dataset