blueoil.networks.segmentation.lm_segnet_v0 Module Contents Classes


LM customized SegNet Network.


LM customized Segnet quantize network.

class blueoil.networks.segmentation.lm_segnet_v0.LmSegnetV0(*args, **kwargs)

Bases: blueoil.networks.segmentation.base.SegnetBase

LM customized SegNet Network.

_get_lmnet_block(self, is_training, channels_data_format)
_max_pool_with_argmax(self, inputs=None, ksize=None, strides=None, padding=None, name='')
_unpool_with_argmax(self, inputs=None, mask=None, ksize=None, name='')
base(self, images, is_training, *args, **kwargs)

Base function contains inference.

  • images – Input images.

  • is_training – A flag for if is training.


Inference result.

Return type


class blueoil.networks.segmentation.lm_segnet_v0.LmSegnetV0Quantize(activation_quantizer=None, activation_quantizer_kwargs={}, weight_quantizer=None, weight_quantizer_kwargs={}, *args, **kwargs)

Bases: blueoil.networks.segmentation.lm_segnet_v0.LmSegnetV0

LM customized Segnet quantize network.

Following args are used for inference: activation_quantizer, activation_quantizer_kwargs, weight_quantizer, weight_quantizer_kwargs.

  • activation_quantizer (callable) – Weight quantizater. See more at blueoil.quantizations.

  • activation_quantizer_kwargs (dict) – Kwargs for activation_quantizer.

  • weight_quantizer (callable) – Activation quantizater. See more at blueoil.quantizations.

  • weight_quantizer_kwargs (dict) – Kwargs for weight_quantizer.

static _quantized_variable_getter(getter, name, weight_quantization=None, *args, **kwargs)

Get the quantized variables.

Use if to choose or skip the target should be quantized.

  • getter – Default from tensorflow.

  • name – Default from tensorflow.

  • weight_quantization – Callable object which quantize variable.

  • args – Args.

  • kwargs – Kwargs.