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.

Inverted residual block structure of MobileNetV2.

More »

Fig 1 Expand

Fig 2.

PRNET [6].

More »

Fig 2 Expand

Fig 3.

Data parallelism.

More »

Fig 3 Expand

Fig 4.

Model parallelism.

More »

Fig 4 Expand

Fig 5.

Cropping and aligning image.

More »

Fig 5 Expand

Fig 6.

Generating depth image from RGB image.

More »

Fig 6 Expand

Fig 7.

Example of horizontal flipping.

More »

Fig 7 Expand

Fig 8.

RGBD input layer.

More »

Fig 8 Expand

Fig 9.

RGB+D input layer.

More »

Fig 9 Expand

Fig 10.

Example of extending the block (1).

More »

Fig 10 Expand

Fig 11.

Example of extending the block (2).

More »

Fig 11 Expand

Fig 12.

Example of weight transferring.

More »

Fig 12 Expand

Fig 13.

Example of the workflow of automatic model finding on distributed training.

More »

Fig 13 Expand

Fig 14.

Example of automatic model finding on distributed training.

More »

Fig 14 Expand

Fig 15.

Example of the workflow of the automatic model finding on concurrent training.

More »

Fig 15 Expand

Table 1.

Performance comparison between each model condition.

More »

Table 1 Expand

Fig 16.

Result of automatic model finding on distributed training.

More »

Fig 16 Expand

Fig 17.

Result of the best layer replication positions.

More »

Fig 17 Expand

Fig 18.

Results of the best layer replication position in Server1.

More »

Fig 18 Expand

Fig 19.

Results of the best layer replication position in Server2.

More »

Fig 19 Expand

Table 2.

Performance comparison between distributed training and concurrent training.

More »

Table 2 Expand