if i want to use "Learnable Ophthalmology SAM" model, how to edit the json file.

[sym330]

If i want to use train "Learnable Ophthalmology SAM" model on ciiyscape dataset, how to edit the json file?
{
"init_lr" : 0.01,
"momentum" : 0.9,
"weight_decay": 5e-4,
"root" : "D:\workspace\datasets\segmentation\CamVid\preprocessed",
"epochs": 300,
"lr_sche": "poly",
"image_size": [360, 480],
"super_reso": true,
"fusion": true,
"num_classes": 32,
"gpu_index": "0",
"model_name": "deeplabv3plus",
"backbone": "resnet101",
"channel": 3,
"upscale_rate": 2,
"num_workers": 4,
"ckpt_dir": "./ckpt",
"dataset": "camvid",
"batch_size": 2
}

[Qsingle]

Thank you for your attention. You need to modify the training script to add the model. We suggest you to modify the scripts/cityscape_train.py.
The following is the script to import the model to the code.

from imed_vision.models.segmentation.learnable_sam import PromptSAM

Then, you need to add the following codes after line 119.

    elif model_name == "promptsam":
         model = PromptSAM(num_classes=num_classes, model_name=backbone, checkpoint=config['pretrained'])

Note: You need to add the parameter pretrained to the JSON config file.