learnt-scattering/main.py at main · BonnerLab/learnt-scattering · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
import os
import shutil
import time
import sys

import numpy as np
import torch
import torch.nn as nn
import torch.nn.parallel
import torch.backends.cudnn as cudnn
import torch.optim
import torch.utils.data

from scattering import Scattering2D, ScatNonLinearity, ScatNonLinearityAndSkip, Realifier, Complexifier
from projection import ComplexConv2d
from classifier import Classifier
from standardization import Standardization, Normalization

from utils import *


def build_layers(input_type: SplitTensorType, modules, psi_modules=[], i: int = 0, num_channels: List[int] = []):
    """ Builds a list of layers from a list of modules.
    :param input_type:
    :param modules: list of strings describing the layers
    :poaram psi_modules: list of strings describing the layers applied after the high-frequency filters
    :param i: index of the current block
    :param num_channels: number of output channels for each block
    :return: Sequential module
    """
    builder = Builder(input_type)

    def branching_kwargs(**submodules):
        """ Builds kwargs for Branching. Expects a dict of module_name -> architecture list of strings. """
        kwargs = {}
        for name, arch in submodules.items():
            kwargs[f"{name}_module_class"] = build_layers
            kwargs[f"{name}_module_kwargs"] = dict(modules=arch, i=i)
        return kwargs

    for module in modules:
        if module == "Fw":
            kwargs = dict(
                scales_per_octave=1, L=8, full_angles=False,
            )
            two_blocks_per_scale_after_block = len(num_channels) % 2
            if i >= two_blocks_per_scale_after_block:
                kwargs.update(factorize_filters=True, i=(i - two_blocks_per_scale_after_block) % 2)

            builder.add_layer(Scattering2D, kwargs)

            kwargs = dict(phi=[], psi=psi_modules)
            builder.add_layer(Branching, branching_kwargs(**kwargs))

        elif module == "R":
            builder.add_batched(Realifier)

        elif module == "C":
            builder.add_batched(Complexifier)

        elif module in ["mod", "rho"]:
            kwargs = dict(
                non_linearity="mod", bias=None, gain=None, learned_params=False,
            )
            if module == "mod":
                builder.add_layer(ScatNonLinearity, kwargs)
            elif module == "rho":
                builder.add_layer(ScatNonLinearityAndSkip, kwargs)
                builder.add_layer(Branching, branching_kwargs(linear=[], non_linear=[]))  # Possibility to add different modules to the linear/non_linear part

        elif module == "Std":
            builder.add_batched(Standardization, dict(remove_mean=True))

        elif module in ["P", "Pr", "Pc"]:
            out_channels = {0: num_channels[i]}
            # Determine type of weights (default is type of input).
            complex_weights = dict(P=None, Pr=True, Pc=False)[module]

            kwargs = dict(
                complex_weights=complex_weights, out_channels=out_channels,
            )

            builder.add_diagonal(ComplexConv2d, kwargs)

        elif module == "N":
            builder.add_diagonal(Normalization)

        elif module == "id":
            builder.add_layer(Identity)

        else:
            assert False

    return builder.module()


def load_model():
    skip = False  # Whether to include a skip-connection over the complex modulus or not
    if skip:
        arch = ["Fw", "rho", "Std", "P", "N"]
        psi_arch = []
    else:
        arch = ["Fw", "Std", "P", "N"]
        psi_arch = ["mod"]
    dataset = "imagenet"

    num_channels_in = 3
    if dataset == "imagenet":
        num_channels = [32, 64, 64, 128, 256, 512, 512, 512, 512, 512, 256]
        spatial_size = 224
        num_classes = 1000
    elif dataset == "cifar":
        num_channels = [64, 128, 256, 512, 512, 512, 512, 512]
        spatial_size = 32
        num_classes = 10
    num_blocks = len(num_channels)

    input_type = TensorType(num_channels=num_channels_in, spatial_shape=(spatial_size, spatial_size), complex=False)
    builder = Builder(input_type)

    builder.add_layer(ToSplitTensor, dict(groups={(0): num_channels_in}))
    for i in range(num_blocks):
        builder.add_layer(build_layers, dict(modules=arch, psi_modules=psi_arch, i=i, num_channels=num_channels))
    builder.add_layer(ToTensor)

    builder.add_layer(
        Classifier, dict(
            nb_classes=num_classes, avg_ker_size=1, avgpool=False, bias=True, batch_norm="affine",
        ),
    )

    model = builder.module()

    return model


if __name__ == '__main__':
    model = load_model()
    print(model)
    print(model(torch.randn(1, 3, 224, 224)))