Kindle

Kindle Modules

Supported Modules Summary

Module Components Arguments
Conv Conv -> BatchNorm -> Activation [out_channels, kernel_size, stride, padding, groups, activation]
DWConv DWConv -> BatchNorm -> Activation [out_channels, kernel_size, stride, padding, activation]
Focus Reshape x -> Conv -> Concat [out_channels, kernel_size, stride, padding, activation]
Bottleneck Expansion ConvBNAct -> ConvBNAct [out_channels, shortcut, groups, expansion, activation]
BottleneckCSP CSP Bottleneck [out_channels, shortcut, groups, expansion, activation]
C3 CSP Bottleneck with 3 Conv [out_channels, shortcut, groups, expansion, activation]
MV2Block MobileNet v2 block [out_channels, stride, expand_ratio, activation]
AvgPool Average pooling [kernel_size, stride, padding]
MaxPool Max pooling [kernel_size, stride, padding]
GlobalAvgPool Global Average Pooling []
SPP Spatial Pyramid Pooling [out_channels, [kernel_size1, kernel_size2, ...], activation]
SPPF Spatial Pyramid Pooling - Fast [out_channels, kernel_size, activation]
Flatten Flatten []
Concat Concatenation [dimension]
Linear Linear [out_channels, activation]
Add Add []
UpSample UpSample []
Identity Identity []
YamlModule Custom module from yaml file ['yaml/file/path', arg0, arg1, ...]
nn.{module_name} PyTorch torch.nn.* module Please refer to https://pytorch.org/docs/stable/nn.html
Pretrained timm.create_model [model_name, use_feature_maps, features_only, pretrained]
PreTrainedFeatureMap Bypass feature layer map from Pretrained [feature_idx]
YOLOHead YOLOv5 head module [n_classes, anchors, out_xyxy]
MobileViTBlock MobileVit Block(experimental) [conv_channels, mlp_channels, depth, kernel_size, patch_size, dropout, activation]

Note

nn.{module_name} is currently experimental. This might change in the future release. Use with caution.

Conv

Argument name Type Default value Description
out_channels int Conv channels
kernel_size int (n, n) kernel size
stride int 1 Conv stride
padding int None Conv padding. If None, auto-padding will be applied which generates same width and height of the input
groups int 1 Group convolution size. If 1, no group convolution
activation str or None "ReLU" If None, no activation(Identity) is applied.

DWConv

Argument name Type Default value Description
out_channels int Conv channels
kernel_size int (n, n) kernel size
stride int 1 Conv stride
padding int None Conv padding. If None, auto-padding will be applied which generates same width and height of the input
activation str or None "ReLU" If None, no activation(Identity) is applied.
  • DWConv is identical to Conv but with force grouped convolution.

Focus

Argument name Type Default value Description
out_channels int Conv channels
kernel_size int (n, n) kernel size
stride int 1 Conv stride
padding int None Conv padding. If None, auto-padding will be applied which generates same width and height of the input
groups int 1 Group convolution size. If 1, no group convolution
activation str or None "ReLU" If None, no activation(Identity) is applied.

Bottleneck

Argument name Type Default value Description
out_channels int Conv channels
shortcut bool True Use shortcut. Only applied when in_channels and out_channels are same.
groups int 1 Group convolution size. If 1, no group convolution
expansion float 0.5 Expansion(squeeze) ratio.
activation str or None "ReLU" If None, no activation(Identity) is applied.

BottleneckCSP

Argument name Type Default value Description
out_channels int Conv channels
shortcut bool True Use shortcut. Only applied when in_channels and out_channels are same.
groups int 1 Group convolution size. If 1, no group convolution
expansion float 0.5 Expansion(squeeze) ratio.
activation str or None "ReLU" If None, no activation(Identity) is applied.

C3

Argument name Type Default value Description
out_channels int Conv channels
shortcut bool True Use shortcut. Only applied when in_channels and out_channels are same.
groups int 1 Group convolution size. If 1, no group convolution
expansion int 0.5 Expansion(squeeze) ratio.
activation str or None "ReLU" If None, no activation(Identity) is applied.

MV2Block

Argument name Type Default value Description
out_channels int Conv channels
stride int 1 Stride value. (1 or 2 only)
expand_ratio int 4 Expansion ratio.
activation str or None "ReLU" If None, no activation(Identity) is applied.

AvgPool

Argument name Type Default value Description
kernel_size int
stride int or None None
padding int 0
ceil_mode bool False
count_include_pad bool True
divisor_override bool or None None

MaxPool

Argument name Type Default value Description
kernel_size int
stride int or None None
padding int 0
dilation int 1
return_indices bool False
ceil_mode bool False

SPP

Argument name Type Default value Description
out_channels int Conv channels
kernel_sizes List[int] [5, 9, 13] List of (n, n) kernel size
activation str or None "ReLU" If None, no activation(Identity) is applied.

SPPF

Argument name Type Default value Description
out_channels int Conv channels
kernel_sizes int 5 Kernel size. Default value(5) is equivalent (5, 9, 13) kernel sizes in SPP
activation str or None "ReLU" If None, no activation(Identity) is applied.

Flatten

Argument name Type Default value Description
start_dim int 1
end_dim int -1

Concat

Argument name Type Default value Description
dimension int 1

Linear

Argument name Type Default value Description
out_channels int
activation str or None None

UpSample

Argument name Type Default value Description
size int or None None
scale_factor int or None 2
mode str nearest
align_corners bool or None None

YamlModule

Argument name Type Default value Description
verbose bool False
  • yaml file path and argument configured in yaml module can not be passed through keyword argument.

Pretrained

Argument name Type Default value Description
model_name str Please refer to https://rwightman.github.io/pytorch-image-models/results for supported models.
ues_feature_maps bool False If True, return value of the module will be list of each feature maps. List[torch.Tensor] (features_only must be True in this case). Otherwise, returns last feature map.
features_only bool True If True, skip classification layer and use feature layers only. torch.Tensor
pretrained bool True use pretrained weight
In case you are just lazy as I am, here is list of pretrained model names. (timm==0.4.5)

'adv_inception_v3', 'cspdarknet53', 'cspresnet50', 'cspresnext50', 'densenet121', 'densenet161', 'densenet169', 'densenet201', 'densenetblur121d', 'dla34', 'dla46_c', 'dla46x_c', 'dla60', 'dla60_res2net', 'dla60_res2next', 'dla60x', 'dla60x_c', 'dla102', 'dla102x', 'dla102x2', 'dla169', 'dm_nfnet_f0', 'dm_nfnet_f1', 'dm_nfnet_f2', 'dm_nfnet_f3', 'dm_nfnet_f4', 'dm_nfnet_f5', 'dm_nfnet_f6', 'dpn68', 'dpn68b', 'dpn92', 'dpn98', 'dpn107', 'dpn131', 'ecaresnet26t', 'ecaresnet50d', 'ecaresnet50d_pruned', 'ecaresnet50t', 'ecaresnet101d', 'ecaresnet101d_pruned', 'ecaresnet269d', 'ecaresnetlight', 'efficientnet_b0', 'efficientnet_b1', 'efficientnet_b1_pruned', 'efficientnet_b2', 'efficientnet_b2_pruned', 'efficientnet_b2a', 'efficientnet_b3', 'efficientnet_b3_pruned', 'efficientnet_b3a', 'efficientnet_em', 'efficientnet_es', 'efficientnet_lite0', 'ens_adv_inception_resnet_v2', 'ese_vovnet19b_dw', 'ese_vovnet39b', 'fbnetc_100', 'gernet_l', 'gernet_m', 'gernet_s', 'gluon_inception_v3', 'gluon_resnet18_v1b', 'gluon_resnet34_v1b', 'gluon_resnet50_v1b', 'gluon_resnet50_v1c', 'gluon_resnet50_v1d', 'gluon_resnet50_v1s', 'gluon_resnet101_v1b', 'gluon_resnet101_v1c', 'gluon_resnet101_v1d', 'gluon_resnet101_v1s', 'gluon_resnet152_v1b', 'gluon_resnet152_v1c', 'gluon_resnet152_v1d', 'gluon_resnet152_v1s', 'gluon_resnext50_32x4d', 'gluon_resnext101_32x4d', 'gluon_resnext101_64x4d', 'gluon_senet154', 'gluon_seresnext50_32x4d', 'gluon_seresnext101_32x4d', 'gluon_seresnext101_64x4d', 'gluon_xception65', 'hrnet_w18', 'hrnet_w18_small', 'hrnet_w18_small_v2', 'hrnet_w30', 'hrnet_w32', 'hrnet_w40', 'hrnet_w44', 'hrnet_w48', 'hrnet_w64', 'ig_resnext101_32x8d', 'ig_resnext101_32x16d', 'ig_resnext101_32x32d', 'ig_resnext101_32x48d', 'inception_resnet_v2', 'inception_v3', 'inception_v4', 'legacy_senet154', 'legacy_seresnet18', 'legacy_seresnet34', 'legacy_seresnet50', 'legacy_seresnet101', 'legacy_seresnet152', 'legacy_seresnext26_32x4d', 'legacy_seresnext50_32x4d', 'legacy_seresnext101_32x4d', 'mixnet_l', 'mixnet_m', 'mixnet_s', 'mixnet_xl', 'mnasnet_100', 'mobilenetv2_100', 'mobilenetv2_110d', 'mobilenetv2_120d', 'mobilenetv2_140', 'mobilenetv3_large_100', 'mobilenetv3_rw', 'nasnetalarge', 'nf_regnet_b1', 'nf_resnet50', 'nfnet_l0c', 'pnasnet5large', 'regnetx_002', 'regnetx_004', 'regnetx_006', 'regnetx_008', 'regnetx_016', 'regnetx_032', 'regnetx_040', 'regnetx_064', 'regnetx_080', 'regnetx_120', 'regnetx_160', 'regnetx_320', 'regnety_002', 'regnety_004', 'regnety_006', 'regnety_008', 'regnety_016', 'regnety_032', 'regnety_040', 'regnety_064', 'regnety_080', 'regnety_120', 'regnety_160', 'regnety_320', 'repvgg_a2', 'repvgg_b0', 'repvgg_b1', 'repvgg_b1g4', 'repvgg_b2', 'repvgg_b2g4', 'repvgg_b3', 'repvgg_b3g4', 'res2net50_14w_8s', 'res2net50_26w_4s', 'res2net50_26w_6s', 'res2net50_26w_8s', 'res2net50_48w_2s', 'res2net101_26w_4s', 'res2next50', 'resnest14d', 'resnest26d', 'resnest50d', 'resnest50d_1s4x24d', 'resnest50d_4s2x40d', 'resnest101e', 'resnest200e', 'resnest269e', 'resnet18', 'resnet18d', 'resnet26', 'resnet26d', 'resnet34', 'resnet34d', 'resnet50', 'resnet50d', 'resnet101d', 'resnet152d', 'resnet200d', 'resnetblur50', 'resnetv2_50x1_bitm', 'resnetv2_50x1_bitm_in21k', 'resnetv2_50x3_bitm', 'resnetv2_50x3_bitm_in21k', 'resnetv2_101x1_bitm', 'resnetv2_101x1_bitm_in21k', 'resnetv2_101x3_bitm', 'resnetv2_101x3_bitm_in21k', 'resnetv2_152x2_bitm', 'resnetv2_152x2_bitm_in21k', 'resnetv2_152x4_bitm', 'resnetv2_152x4_bitm_in21k', 'resnext50_32x4d', 'resnext50d_32x4d', 'resnext101_32x8d', 'rexnet_100', 'rexnet_130', 'rexnet_150', 'rexnet_200', 'selecsls42b', 'selecsls60', 'selecsls60b', 'semnasnet_100', 'seresnet50', 'seresnet152d', 'seresnext26d_32x4d', 'seresnext26t_32x4d', 'seresnext50_32x4d', 'skresnet18', 'skresnet34', 'skresnext50_32x4d', 'spnasnet_100', 'ssl_resnet18', 'ssl_resnet50', 'ssl_resnext50_32x4d', 'ssl_resnext101_32x4d', 'ssl_resnext101_32x8d', 'ssl_resnext101_32x16d', 'swsl_resnet18', 'swsl_resnet50', 'swsl_resnext50_32x4d', 'swsl_resnext101_32x4d', 'swsl_resnext101_32x8d', 'swsl_resnext101_32x16d', 'tf_efficientnet_b0', 'tf_efficientnet_b0_ap', 'tf_efficientnet_b0_ns', 'tf_efficientnet_b1', 'tf_efficientnet_b1_ap', 'tf_efficientnet_b1_ns', 'tf_efficientnet_b2', 'tf_efficientnet_b2_ap', 'tf_efficientnet_b2_ns', 'tf_efficientnet_b3', 'tf_efficientnet_b3_ap', 'tf_efficientnet_b3_ns', 'tf_efficientnet_b4', 'tf_efficientnet_b4_ap', 'tf_efficientnet_b4_ns', 'tf_efficientnet_b5', 'tf_efficientnet_b5_ap', 'tf_efficientnet_b5_ns', 'tf_efficientnet_b6', 'tf_efficientnet_b6_ap', 'tf_efficientnet_b6_ns', 'tf_efficientnet_b7', 'tf_efficientnet_b7_ap', 'tf_efficientnet_b7_ns', 'tf_efficientnet_b8', 'tf_efficientnet_b8_ap', 'tf_efficientnet_cc_b0_4e', 'tf_efficientnet_cc_b0_8e', 'tf_efficientnet_cc_b1_8e', 'tf_efficientnet_el', 'tf_efficientnet_em', 'tf_efficientnet_es', 'tf_efficientnet_l2_ns', 'tf_efficientnet_l2_ns_475', 'tf_efficientnet_lite0', 'tf_efficientnet_lite1', 'tf_efficientnet_lite2', 'tf_efficientnet_lite3', 'tf_efficientnet_lite4', 'tf_inception_v3', 'tf_mixnet_l', 'tf_mixnet_m', 'tf_mixnet_s', 'tf_mobilenetv3_large_075', 'tf_mobilenetv3_large_100', 'tf_mobilenetv3_large_minimal_100', 'tf_mobilenetv3_small_075', 'tf_mobilenetv3_small_100', 'tf_mobilenetv3_small_minimal_100', 'tresnet_l', 'tresnet_l_448', 'tresnet_m', 'tresnet_m_448', 'tresnet_xl', 'tresnet_xl_448', 'tv_densenet121', 'tv_resnet34', 'tv_resnet50', 'tv_resnet101', 'tv_resnet152', 'tv_resnext50_32x4d', 'vgg11', 'vgg11_bn', 'vgg13', 'vgg13_bn', 'vgg16', 'vgg16_bn', 'vgg19', 'vgg19_bn', 'vit_base_patch16_224', 'vit_base_patch16_224_in21k', 'vit_base_patch16_384', 'vit_base_patch32_224_in21k', 'vit_base_patch32_384', 'vit_base_resnet50_224_in21k', 'vit_base_resnet50_384', 'vit_deit_base_distilled_patch16_224', 'vit_deit_base_distilled_patch16_384', 'vit_deit_base_patch16_224', 'vit_deit_base_patch16_384', 'vit_deit_small_distilled_patch16_224', 'vit_deit_small_patch16_224', 'vit_deit_tiny_distilled_patch16_224', 'vit_deit_tiny_patch16_224', 'vit_large_patch16_224', 'vit_large_patch16_224_in21k', 'vit_large_patch16_384', 'vit_large_patch32_224_in21k', 'vit_large_patch32_384', 'vit_small_patch16_224', 'wide_resnet50_2', 'wide_resnet101_2', 'xception', 'xception41', 'xception65', 'xception71'

PretrainedFeatureMap

Argument name Type Default value Description
feature_idx int -1 Index of the feature maps

YOLOHead

Argument name Type Default value Description
n_classes int Number of classes to detect
anchors List[List[float]] Anchor lists. Each list represents each layer's anchor and each components in the list represents anchor size of [w1, h1, w2, h2, ...]
out_xyxy bool False Return coordinates as xyxy format. (For older version of yolov5 compatability)

Note

[[-3, -2, -1], 1, YOLOHead, [80, [[100, 200, 200, 100, 200, 200], [50, 100, 100, 50, 100, 100], [10, 20, 20, 10, 20, 20]]]]

This represents that YOLOHead takes inputs from previous 3 layers, detects 80 classes with 3 layers of 3 anchors.

MobileViTBlock

Argument name Type Default value Description
conv_channels int number of channels in convolution
mlp_channels int number of channels in MLP
depth int depth of the transformer
kernel_size int 3 (n, n) kernel size
patch_size int or tuple 2 patch size to use in transformer.
dropout float 0.0 Dropout probability used in Attention and MLP
activation str or None SiLU If None, no activation(Identity) is applied.