from __future__ import print_function import os, sys, inspect currentdir = os.path.dirname( os.path.abspath(inspect.getfile(inspect.currentframe()))) parentdir = os.path.dirname(os.path.dirname(currentdir)) sys.path.insert(0, parentdir) from utils.args import args import setup.categories.ae_setup as AESetup from models.autoencoders import * from datasets.DRD import DRD if __name__ == "__main__": dataset = DRD(root_path=os.path.join(args.root_path, "diabetic-retinopathy-detection"), downsample=64) model = ALILikeVAE(dims=(3, 64, 64)) AESetup.train_variational_autoencoder(args, model=model, dataset=dataset.get_D1_train(), BCE_Loss=False)
from __future__ import print_function import os,sys,inspect currentdir = os.path.dirname(os.path.abspath(inspect.getfile(inspect.currentframe()))) parentdir = os.path.dirname(os.path.dirname(currentdir)) sys.path.insert(0,parentdir) from utils.args import args import setup.categories.ae_setup as AESetup from models.autoencoders import * from datasets.DRD import DRD if __name__ == "__main__": dataset = DRD(root_path=os.path.join(args.root_path, "diabetic-retinopathy-detection"), downsample=64) model = ALILikeVAE(dims=(3, 64, 64)) AESetup.train_variational_autoencoder(args, model=model, dataset=dataset.get_D1_train(), BCE_Loss=True)
import global_vars as Global from utils.args import args import torchvision from torch.utils.data.sampler import WeightedRandomSampler import numpy as np from torch import optim import setup.categories.classifier_setup as CLSetup from models.classifiers import DRDDense from datasets.DRD import DRD class DRDDenseCustom(DRDDense): def train_config(self): config = {} if self.train_features: config['optim'] = optim.Adam( [{'params':self.densenet121.classifier.parameters(), 'lr':1e-3}, {'params':self.densenet121.features.parameters()}], lr=1e-3) else: config['optim'] = optim.Adam(self.densenet121.classifier.parameters(), lr=1e-3, ) config['scheduler'] = optim.lr_scheduler.StepLR(config['optim'], 30, gamma=0.5) config['max_epoch'] = 100 return config if __name__ == "__main__": dataset = DRD(root_path=os.path.join(args.root_path, "diabetic-retinopathy-detection")) model = DRDDenseCustom(train_features=True, pretrained_weights_path="./model.pth.tar") args.num_classes = 2 CLSetup.train_classifier(args, model=model, dataset=dataset.get_D1_train(), balanced=True)