lingvo.tasks.car.car_layers module

Common layers for car models.

These are usually sub-classes of base_layer.BaseLayer used by builder_lib.

class lingvo.tasks.car.car_layers.SamplingAndGroupingLayer(*args, **kwargs)[source]

Bases: lingvo.core.base_layer.BaseLayer

Sampling and Grouping layer (based on PointNet++).

classmethod Params()[source]

Returns the layer params.

FProp(theta, input_data)[source]

Apply projection to inputs.

Parameters

  • theta – A NestedMap object containing weights’ values of this layer and its children layers.

  • input_data – A NestedMap object containing ‘points’, ‘features’, ‘padding’ Tensors, all of type tf.float32. ‘points’: Shape [N, P1, 3] ‘features’: Shape [N, P1, F] ‘padding’: Shape [N, P1] where 0 indicates real, 1 indicates padded.

Returns

A NestedMap consisting of the following two NestedMaps,

grouped_points: consists of the grouped points, features and padding. query_points: consists of the sampled points and padding.