Segmentation Post-processing
Base class: SemanticSegPostProcessBase
Supported Model Post-processing
Model Post-processing |
Description |
|---|---|
Deeplabv3PostProcess |
Deeplabv3 post-processing |
Deeplabv3PytorchPost |
Deeplabv3 PyTorch post-processing |
Deeplabv3TensorflowPost |
Deeplabv3 TensorFlow post-processing |
Deeplabv3MindsporePost |
Deeplabv3 MindSpore post-processing |
UNetMindSporePostProcess |
UNet MindSpore post-processing |
Unified Constructor
postProcessName(config_data, label_path)
postProcessName(config_path, label_path)
Example: Deeplabv3PostProcess (config_data, label_path) or Deeplabv3PostProcess(config_path, label_path)
Input Parameter of the Constructor
Parameter |
Type |
Description |
|---|---|---|
config_data |
dict |
Configuration data |
label_path |
str |
Label path |
config_path |
str |
Configuration path |
Output Parameter of the Constructor
Corresponding postprocessing object
Unified Post-processing Function
process(tensor, params)
process(tensor, param)
Input Parameter Description of the Post-processing Function
Parameter |
Type |
Description |
|---|---|---|
tensor |
List[base.Tensor] |
Tensor list |
params |
List[base.ResizedImageInfo] |
ResizedImageInfo list |
param |
base.ResizedImageInfo |
ResizedImageInfo object |
Output Parameter Description of the Post-processing Function
list[SemanticSegInfo]
Exception Throwing Interface
If the tensor fails to be passed and copied to the host, the runtime exception is thrown.