8.1.1.7.1.2.1.2. blueoil.networks.keypoint_detection.lm_single_pose_v1

8.1.1.7.1.2.1.2.1. Module Contents

8.1.1.7.1.2.1.2.1.1. Classes

LmSinglePoseV1

LM original single-person pose estimation network.

LmSinglePoseV1Quantize

LM original quantized single-person pose estimation network.

class blueoil.networks.keypoint_detection.lm_single_pose_v1.LmSinglePoseV1(stride=2, *args, **kwargs)

Bases: blueoil.networks.keypoint_detection.base.Base

LM original single-person pose estimation network.

loss(self, output, labels)

Loss function for single-person pose estimation.

Parameters
  • output – Tensor of network output. shape is (batch_size, height, width, num_joints).

  • labels – Tensor of ground-truth labels. shape is (batch_size, height, width, num_joints).

_get_lmnet_block(self, is_training, channels_data_format)
_space_to_depth(self, inputs=None, block_size=2, name='')
base(self, images, is_training, *args, **kwargs)

Base function contains inference.

Parameters
  • images – Input images.

  • is_training – A flag for if is training.

Returns

Inference result.

Return type

tf.Tensor

class blueoil.networks.keypoint_detection.lm_single_pose_v1.LmSinglePoseV1Quantize(activation_quantizer=None, activation_quantizer_kwargs={}, weight_quantizer=None, weight_quantizer_kwargs={}, *args, **kwargs)

Bases: blueoil.networks.keypoint_detection.lm_single_pose_v1.LmSinglePoseV1

LM original quantized single-person pose estimation network.

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

Parameters
  • 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.

Parameters
  • getter – Default from tensorflow.

  • name – Default from tensorflow.

  • weight_quantization – Callable object which quantize variable.

  • args – Args.

  • kwargs – Kwargs.