Loading Pre-trained custom model

[ishangupta3]

I am following the project structure and the custom.yaml file, I would like to load the pre trained model myself. Before loading the model, I would need to instantiate the class like this below:

model = TheModelClass(*args, **kwargs)

According to the custom.yaml file; the 'task' is pvnet, but the network is res.
clean-pvnet/lib/networks/pvnet

Which class network would the appropriate one to use when loading the custom model in pytorch?

Would it be this one: Resnet18?
https://github.com/zju3dv/clean-pvnet/blob/07e7c09e80cde938e013d8bc2880605add33a0ce/lib/networks/pvnet/resnet18.py

Or this one: PoseResNet?
https://github.com/zju3dv/clean-pvnet/blob/07e7c09e80cde938e013d8bc2880605add33a0ce/lib/networks/resnet.py

My end goal is to just take my pretrained model that I trained; and create a brand new inference code pipeline, where I can customize it with my own usecase.

[pengsida]

If you want to train the pvnet on your custom data, please follow these steps: https://github.com/zju3dv/clean-pvnet#training-on-the-custom-object

[ishangupta3]

I already have the custom trained model already trained.

I would like to load it myself, I was asking which network class I should use to load the model.

[ishangupta3]

Pytorch requires you to set up model in this format: (I already have the model trained on my own custom dataset) I would like to load the model myself.

model = TheModelClass(*args, **kwargs)
model.load_state_dict(torch.load(PATH))

[pengsida]

python run.py --type visualize --cfg_file configs/custom.yaml
The network is defined in configs/custom.yaml.
The network is https://github.com/zju3dv/clean-pvnet/blob/master/lib/networks/pvnet/__init__.py#L5