Util/Custom_Model.py

import timm
import torch.nn as nn

class Build_Custom_Model(nn.Module):
    def __init__(self, model_name, target_size, pretrained=False):
        super().__init__()
        self.model = timm.create_model(model_name, pretrained=pretrained, in_chans=1)
        if(model_name=="vit_base_patch16_224" or model_name=="swin_base_patch4_window7_224"):
            self.n_features = self.model.head.in_features
            self.model.head = nn.Linear(self.n_features, target_size)
        if(model_name=="resnet34d"):
            self.n_features = self.model.fc.in_features
            self.model.fc = nn.Linear(self.n_features, target_size)
        if(model_name=="resnet18d"):
            self.n_features = self.model.fc.in_features
            self.model.fc = nn.Linear(self.n_features, target_size)
        if(model_name=="tf_efficientnet_b7_ns"):
            self.n_features = self.model.classifier.in_features
            self.model.classifier = nn.Linear(self.n_features, target_size)
        if(model_name=="tf_efficientnet_b0_ns"):
            self.n_features = self.model.classifier.in_features
            self.model.classifier = nn.Linear(self.n_features, target_size)
        if(model_name=="tf_efficientnet_lite0"):
            self.n_features = self.model.classifier.in_features
            self.model.classifier = nn.Linear(self.n_features, target_size)
        if(model_name=="mobilenetv2_050"):
            self.n_features = self.model.classifier.in_features
            self.model.classifier = nn.Linear(self.n_features, target_size)
        if(model_name=="eca_nfnet_l0"):
            self.n_features = self.model.head.fc.in_features
            self.model.head.fc = nn.Linear(self.n_features, target_size)

    def forward(self, x):
        output = self.model(x)
        return output

def reshape_transform(tensor, height=7, width=7):
    result = tensor.reshape(tensor.size(0),
        height, width, tensor.size(2))

    # Bring the channels to the first dimension,
    # like in CNNs.
    result = result.transpose(2, 3).transpose(1, 2)
    return result