Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

Key technologies of the proposed method.

More »

Fig 1 Expand

Fig 2.

TDCN structure.

More »

Fig 2 Expand

Table 1.

Important notations and the corresponding definitions in this work.

More »

Table 1 Expand

Fig 3.

Edge operators, normal convolutions, and trainable absolute gradient (TAG) layer.

More »

Fig 3 Expand

Fig 4.

Structure of difference convolution module.

More »

Fig 4 Expand

Fig 5.

Typical self-attention layer and proposed boundary-aware attention layer.

More »

Fig 5 Expand

Fig 6.

Transformer with boundary-aware attention.

More »

Fig 6 Expand

Fig 7.

Structure of the head function.

More »

Fig 7 Expand

Fig 8.

Precision-recall curves on BSDS dataset.

More »

Fig 8 Expand

Table 2.

Performances on BSDS.

More »

Table 2 Expand

Table 3.

Performances on NYUD.

More »

Table 3 Expand

Fig 9.

Number of parameters and corresponding performances (ODS) of different structures.

More »

Fig 9 Expand

Fig 10.

Prediction results for BSDS (top), NYUD-RGB (middle), and NYUD-HHA (bottom).

More »

Fig 10 Expand

Table 4.

Performance comparison between unified prediction and dataset-specific prediction.

More »

Table 4 Expand

Fig 11.

Prediction results under complete TDCN and corresponding ablation study after three epochs.

More »

Fig 11 Expand

Table 5.

Ablation study on TAG layer, boundary-aware attention, and boosting strategy after three training epochs.

More »

Table 5 Expand

Table 6.

Parameter sensitivity analysis for TDCN after three training epochs.

More »

Table 6 Expand