def train(): device = torch.device(conf.cuda if torch.cuda.is_available() else "cpu") dataset = Training_Dataset(conf.data_path_train, conf.gaussian_noise_param, conf.crop_img_size) dataset_length = len(dataset) train_loader = DataLoader(dataset, batch_size=4, shuffle=True, num_workers=4) model = UNet(in_channels=conf.img_channel, out_channels=conf.img_channel) criterion = nn.MSELoss() model = model.to(device) optim = Adam(model.parameters(), lr=conf.learning_rate, betas=(0.9, 0.999), eps=1e-8, weight_decay=0, amsgrad=True) scheduler = lr_scheduler.StepLR(optim, step_size=100, gamma=0.5) model.train() print(model) print("Starting Training Loop...") since = time.time() for epoch in range(conf.max_epoch): print('Epoch {}/{}'.format(epoch, conf.max_epoch - 1)) print('-' * 10) running_loss = 0.0 scheduler.step() for batch_idx, (source, target) in enumerate(train_loader): source = source.to(device) target = target.to(device) optim.zero_grad() denoised_source = model(source) loss = criterion(denoised_source, target) loss.backward() optim.step() running_loss += loss.item() * source.size(0) print('Current loss {} and current batch idx {}'.format( loss.item(), batch_idx)) epoch_loss = running_loss / dataset_length print('{} Loss: {:.4f}'.format('current ' + str(epoch), epoch_loss)) if (epoch + 1) % conf.save_per_epoch == 0: save_model(model, epoch + 1) time_elapsed = time.time() - since print('Training complete in {:.0f}m {:.0f}s'.format( time_elapsed // 60, time_elapsed % 60))
out_norm = 15383.0 LE_img = imgread(os.path.join(dir_path_LE, "LE_01.tif")) LE_512 = cropImage(LE_img, IMG_SHAPE[0],IMG_SHAPE[1]) sample_le = {} for le_512 in LE_512: tiles = crop_prepare(le_512, CROP_STEP, IMG_SIZE) for n,img in enumerate(tiles): if n not in sample_le: sample_le[n] = [] img = transform.resize(img,(IMG_SIZE*2, IMG_SIZE*2),preserve_range=True,order=3) sample_le[n].append(img) SNR_model = UNet(n_channels=15, n_classes=15) print("{} paramerters in total".format(sum(x.numel() for x in SNR_model.parameters()))) SNR_model.cuda(cuda) SNR_model.load_state_dict(torch.load(SNR_model_path)) # SNR_model.load_state_dict(torch.load(os.path.join(dir_path,"model","LE_HE_mito","LE_HE_0825.pkl"))) SNR_model.eval() SIM_UNET = UNet(n_channels=15, n_classes=1) print("{} paramerters in total".format(sum(x.numel() for x in SIM_UNET.parameters()))) SIM_UNET.cuda(cuda) SIM_UNET.load_state_dict(torch.load(SIM_UNET_model_path)) # SIM_UNET.load_state_dict(torch.load(os.path.join(dir_path,"model","HE_HER_mito","HE_X2_HER_0825.pkl"))) SIM_UNET.eval() SRRFDATASET = ReconsDataset( img_dict=sample_le, transform=ToTensor(),
def train(cont=False): # for tensorboard tracking logger = get_logger() logger.info("(1) Initiating Training ... ") logger.info("Training on device: {}".format(device)) writer = SummaryWriter() # init model aux_layers = None if net == "SETR-PUP": aux_layers, model = get_SETR_PUP() elif net == "SETR-MLA": aux_layers, model = get_SETR_MLA() elif net == "TransUNet-Base": model = get_TransUNet_base() elif net == "TransUNet-Large": model = get_TransUNet_large() elif net == "UNet": model = UNet(CLASS_NUM) # prepare dataset cluster_model = get_clustering_model(logger) train_dataset = CityscapeDataset(img_dir=data_dir, img_dim=IMG_DIM, mode="train", cluster_model=cluster_model) valid_dataset = CityscapeDataset(img_dir=data_dir, img_dim=IMG_DIM, mode="val", cluster_model=cluster_model) train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True) valid_loader = DataLoader(valid_dataset, batch_size=batch_size, shuffle=False) logger.info("(2) Dataset Initiated. ") # optimizer epochs = epoch_num if epoch_num > 0 else iteration_num // len( train_loader) + 1 optim = SGD(model.parameters(), lr=lrate, momentum=momentum, weight_decay=wdecay) # optim = Adam(model.parameters(), lr=lrate) scheduler = lr_scheduler.MultiStepLR( optim, milestones=[int(epochs * fine_tune_ratio)], gamma=0.1) cur_epoch = 0 best_loss = float('inf') epochs_since_improvement = 0 # for continue training if cont: model, optim, cur_epoch, best_loss = load_ckpt_continue_training( best_ckpt_src, model, optim, logger) logger.info("Current best loss: {0}".format(best_loss)) with warnings.catch_warnings(): warnings.simplefilter("ignore") for i in range(cur_epoch): scheduler.step() else: model = nn.DataParallel(model) model = model.to(device) logger.info("(3) Model Initiated ... ") logger.info("Training model: {}".format(net) + ". Training Started.") # loss ce_loss = CrossEntropyLoss() if use_dice_loss: dice_loss = DiceLoss(CLASS_NUM) # loop over epochs iter_count = 0 epoch_bar = tqdm.tqdm(total=epochs, desc="Epoch", position=cur_epoch, leave=True) logger.info("Total epochs: {0}. Starting from epoch {1}.".format( epochs, cur_epoch + 1)) for e in range(epochs - cur_epoch): epoch = e + cur_epoch # Training. model.train() trainLossMeter = LossMeter() train_batch_bar = tqdm.tqdm(total=len(train_loader), desc="TrainBatch", position=0, leave=True) for batch_num, (orig_img, mask_img) in enumerate(train_loader): orig_img, mask_img = orig_img.float().to( device), mask_img.float().to(device) if net == "TransUNet-Base" or net == "TransUNet-Large": pred = model(orig_img) elif net == "SETR-PUP" or net == "SETR-MLA": if aux_layers is not None: pred, _ = model(orig_img) else: pred = model(orig_img) elif net == "UNet": pred = model(orig_img) loss_ce = ce_loss(pred, mask_img[:].long()) if use_dice_loss: loss_dice = dice_loss(pred, mask_img, softmax=True) loss = 0.5 * (loss_ce + loss_dice) else: loss = loss_ce # Backward Propagation, Update weight and metrics optim.zero_grad() loss.backward() optim.step() # update learning rate for param_group in optim.param_groups: orig_lr = param_group['lr'] param_group['lr'] = orig_lr * (1.0 - iter_count / iteration_num)**0.9 iter_count += 1 # Update loss trainLossMeter.update(loss.item()) # print status if (batch_num + 1) % print_freq == 0: status = 'Epoch: [{0}][{1}/{2}]\t' \ 'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch+1, batch_num+1, len(train_loader), loss=trainLossMeter) logger.info(status) # log loss to tensorboard if (batch_num + 1) % tensorboard_freq == 0: writer.add_scalar( 'Train_Loss_{0}'.format(tensorboard_freq), trainLossMeter.avg, epoch * (len(train_loader) / tensorboard_freq) + (batch_num + 1) / tensorboard_freq) train_batch_bar.update(1) writer.add_scalar('Train_Loss_epoch', trainLossMeter.avg, epoch) # Validation. model.eval() validLossMeter = LossMeter() valid_batch_bar = tqdm.tqdm(total=len(valid_loader), desc="ValidBatch", position=0, leave=True) with torch.no_grad(): for batch_num, (orig_img, mask_img) in enumerate(valid_loader): orig_img, mask_img = orig_img.float().to( device), mask_img.float().to(device) if net == "TransUNet-Base" or net == "TransUNet-Large": pred = model(orig_img) elif net == "SETR-PUP" or net == "SETR-MLA": if aux_layers is not None: pred, _ = model(orig_img) else: pred = model(orig_img) elif net == "UNet": pred = model(orig_img) loss_ce = ce_loss(pred, mask_img[:].long()) if use_dice_loss: loss_dice = dice_loss(pred, mask_img, softmax=True) loss = 0.5 * (loss_ce + loss_dice) else: loss = loss_ce # Update loss validLossMeter.update(loss.item()) # print status if (batch_num + 1) % print_freq == 0: status = 'Validation: [{0}][{1}/{2}]\t' \ 'Loss {loss.val:.4f} ({loss.avg:.4f})\t'.format(epoch+1, batch_num+1, len(valid_loader), loss=validLossMeter) logger.info(status) # log loss to tensorboard if (batch_num + 1) % tensorboard_freq == 0: writer.add_scalar( 'Valid_Loss_{0}'.format(tensorboard_freq), validLossMeter.avg, epoch * (len(valid_loader) / tensorboard_freq) + (batch_num + 1) / tensorboard_freq) valid_batch_bar.update(1) valid_loss = validLossMeter.avg writer.add_scalar('Valid_Loss_epoch', valid_loss, epoch) logger.info("Validation Loss of epoch [{0}/{1}]: {2}\n".format( epoch + 1, epochs, valid_loss)) # update optim scheduler scheduler.step() # save checkpoint is_best = valid_loss < best_loss best_loss_tmp = min(valid_loss, best_loss) if not is_best: epochs_since_improvement += 1 logger.info("Epochs since last improvement: %d\n" % (epochs_since_improvement, )) if epochs_since_improvement == early_stop_tolerance: break # early stopping. else: epochs_since_improvement = 0 state = { 'epoch': epoch, 'loss': best_loss_tmp, 'model_state_dict': model.state_dict(), 'optimizer_state_dict': optim.state_dict(), } torch.save(state, ckpt_src) logger.info("Checkpoint updated.") best_loss = best_loss_tmp epoch_bar.update(1) writer.close()
batch_size=4, shuffle=True, num_workers=4) classes = ('Buildings', 'MiscMan-made', 'Road', 'Track', 'Trees', 'Crops', 'Waterway', 'Standing_Water', 'Vehicle_Large', 'Vehicle_Small') # Model definition model = UNet(n_classes=len(classes), in_channels=_NUM_CHANNELS_) if torch.cuda.device_count() >= 1: print("Training model on ", torch.cuda.device_count(), "GPUs!") model = nn.DataParallel(model) # Loss function and Optimizer definitions criterion = nn.BCELoss() optimizer = optim.SGD(model.parameters(), lr=0.001, momentum=0.9) # Network training epoch_data = {} for epoch in range(_NUM_EPOCHS_): epoch_loss = 0.0 epoch_data[epoch] = {} for i, data in enumerate(trainloader, 0): # Get the inputs for the network inputs = data['image'].to(_COMPUTE_DEVICE_) labels = data['masks'].to(_COMPUTE_DEVICE_) optimizer.zero_grad() # zero the parameter gradients # Forward pass + Backward pass + Optimisation outputs = model(inputs)
training_dataset = True, in_size = 320, train_in_size = input_size) train_dataloader = torch.utils.data.DataLoader(SRRFDATASET, batch_size=batch_size, shuffle=True, pin_memory=True) # better than for loop SRRFDATASET2 = ReconsDataset(all_data_path="/media/star/LuhongJin/UNC_data/SRRF/New_training_20190829/0NPY_Dataset/Dataset/Microtubule/", maximum_intensity_4normalization_path="/home/star/0_code_lhj/DL-SIM-github/Training_codes/UNetMax_intensity.npy", transform = ToTensor(), training_dataset = False, in_size = 320, train_in_size = input_size) validation_dataloader = torch.utils.data.DataLoader(SRRFDATASET2, batch_size=batch_size, shuffle=True, pin_memory=True) # better than for loop model = UNet(n_channels=input_size, n_classes=output_size) print("{} paramerters in total".format(sum(x.numel() for x in model.parameters()))) model.cuda(cuda) optimizer = torch.optim.Adam(model.parameters(),lr=learning_rate, betas=(0.9, 0.999)) loss_all = np.zeros((2000, 4)) for epoch in range(2000): mae_m, mae_s = val_during_training(train_dataloader) loss_all[epoch,0] = mae_m loss_all[epoch,1] = mae_s mae_m, mae_s = val_during_training(validation_dataloader) loss_all[epoch,2] = mae_m loss_all[epoch,3] = mae_s file = Workbook(encoding = 'utf-8') table = file.add_sheet('loss_all')
#num_train = math.ceil(train_ratio*len(image_dataset)) #num_val = len(image_dataset) - num_train #train_dataset, val_dataset = torch.utils.data.random_split(image_dataset, [num_train, num_val]) # Set up tensor board writer = SummaryWriter(tfboard_dir) # Define a loss function and optimizer weights = torch.ones(num_class) #Give a larger weight to SRLM for SRLM #weights[-1] = 5 weights = weights.to(device) #ignore_index? #criterion = nn.CrossEntropyLoss(weights) criterion = l.GeneralizedDiceLoss(num_classes=num_class, weight=weights) optimizer = optim.Adam(net.parameters(), lr=c.learning_rate) scheduler = optim.lr_scheduler.StepLR(optimizer=optimizer, step_size=c.step_size, gamma=0.1) ## Save and genearte patches for datasets tp_dir = dir_names.patch_dir + "/training_data" vp_dir = dir_names.patch_dir + "/validation_data" #c.force_create(tp_dir) #c.force_create(vp_dir) #if os.path.exists(dir_names.train_patch_csv): # os.remove(dir_names.train_patch_csv) #if os.path.exists(dir_names.val_patch_csv): # os.remove(dir_names.val_patch_csv) #
train_loader = torch.utils.data.DataLoader(X_train, batch_size=batch_size, shuffle=True, pin_memory=False) # better than for loop val_loader = torch.utils.data.DataLoader(y_train, batch_size=batch_size, shuffle=False, pin_memory=False) # better than for loop X_train,y_train,X_test,y_test = None,None,None,None else: train_set = TrainDatasetFromFolder('data/DIV2K_train_HR', crop_size=CROP_SIZE, upscale_factor=UPSCALE_FACTOR) val_set = ValDatasetFromFolder('data/DIV2K_valid_HR', crop_size=CROP_SIZE, upscale_factor=UPSCALE_FACTOR) train_loader = DataLoader(dataset=train_set, num_workers=4, batch_size=64, shuffle=True) val_loader = DataLoader(dataset=val_set, num_workers=4, batch_size=1, shuffle=False) #x,y = next(iter(val_loader)),next(iter(train_loader)) #print(x[0].shape,x[1].shape,x[2].shape,y[0].shape) #netG = Generator(UPSCALE_FACTOR,in_channels,out_channels) netG = UNet(n_channels=in_channels, n_classes=out_channels) #print(summary(netG,(in_channels,128,128))) print('# generator parameters:', sum(param.numel() for param in netG.parameters())) netD = Discriminator(out_channels) print('# discriminator parameters:', sum(param.numel() for param in netD.parameters())) #print(summary(netD,(out_channels,256,256))) generator_criterion = GeneratorLoss() #print(summary(generator_criterion,(3,256,256))) if torch.cuda.is_available(): netG.cuda() netD.cuda() generator_criterion.cuda() optimizerG = optim.Adam(netG.parameters()) optimizerD = optim.Adam(netD.parameters()) results = {'d_loss': [], 'g_loss': [], 'd_score': [], 'g_score': [], 'psnr': [], 'ssim': []}
batch_size = 1 SRRFDATASET = ReconsDataset( test_in_path= "/home/star/0_code_lhj/DL-SIM-github/TESTING_DATA/microtuble/HE_X2/", transform=ToTensor(), img_type='tif', in_size=256) test_dataloader = torch.utils.data.DataLoader( SRRFDATASET, batch_size=batch_size, shuffle=True, pin_memory=True) # better than for loop model = UNet(n_channels=3, n_classes=1) print("{} paramerters in total".format( sum(x.numel() for x in model.parameters()))) model.cuda(cuda) model.load_state_dict( torch.load( "/home/star/0_code_lhj/DL-SIM-github/MODELS/UNet_SIM3_microtubule.pkl" )) model.eval() for batch_idx, items in enumerate(test_dataloader): image = items['image_in'] image_name = items['image_name'] print(image_name[0]) model.train() image = np.swapaxes(image, 1, 3)
n_classes = 2 # instance_model = ReSeg(n_classes=n_classes, use_instance_seg=True, pretrained=False, usegpu=True).to(args.device) segmenter_model = UNet(n_channels=1, n_classes=2).to(args.device) instance_model = UNet(n_channels=1, n_classes=6).to(args.device) loss_binary = torch.nn.BCEWithLogitsLoss() discriminative_loss = DiscriminativeLoss(delta_var=delta_var, delta_dist=delta_dist, norm=2, usegpu=True) cross_entropy_fn = torch.nn.CrossEntropyLoss() optimizer = torch.optim.Adam( list(instance_model.parameters()) + list(segmenter_model.parameters()), 0.001) def stack_iterator(n_portions, block_size, stacks=[]): for y in range(n_portions): for x in range(n_portions): y1 = y * block_size x1 = x * block_size y2 = (y + 1) * block_size x2 = (x + 1) * block_size crops = [s[:, :, y1:y2, x1:x2] for s in stacks] yield crops, (x1, y1, x2, y2)
class PG(object): def __init__(self, configs, env): self.configs = configs self.env = env self.action_size = (64, 1024) # n_channels=3 for RGB images # n_classes is the number of probabilities you want to get per pixel # - For 1 class and background, use n_classes=1 # - For 2 classes, use n_classes=1 # - For N > 2 classes, use n_classes=N # TODO now I assume input<->output size are equal, which might not be true, so we need some modifications onto Unet if necesary self.actor = UNet( n_channels=3, n_classes=1, bilinear=True ) # [B,C, H_in=372, W_in=1242] -> [B, C, H_out=64, W_out=1024] self.optimizer = Adam(self.actor.parameters(), lr=configs['lr']) self.actor.to(device) def get_action(self, state, deterministic=False): """Given the state, produces an action, the probability of the action, the log probability of the action, and the argmax action""" action_probabilities = self.actor( state) # output size should be [B*H*W] action_probabilities = torch.sigmoid( action_probabilities) # make sure the probs are in range [0,1] # B, _, _, _ = action_probabilities.shape action_probabilities = action_probabilities[:, :, :self. action_size[0], :self. action_size[1]] action_probabilities = torch.squeeze(action_probabilities, 1) # assert action_probabilities.size()[1, 2] == self.action_size, "Actor output the wrong size" # action_probabilities_flat = action_probabilities.contiguous().view(B, -1) # TODO leave this to future process; seems it will get the index max_probability_action = torch.argmax(action_probabilities, dim=-1) if deterministic: # using deteministic policy during test time action = action_probabilities(action_probabilities > 0.5).cpu() else: # using stochastic policy during traning time action_distribution = Bernoulli( action_probabilities ) # this creates a distribution to sample from action = action_distribution.sample().cpu( ) # sample the discrete action and copy it to cpu # Have to deal with situation of 0.0 probabilities because we can't do log 0 z = action_probabilities == 0.0 z = z.float() * 1e-8 log_action_probabilities = torch.log(action_probabilities + z) return action, action_probabilities, log_action_probabilities, max_probability_action def compute_loss(self, obs, act, rew): """make loss function whose gradient, for the right data, is policy gradient""" # TODO we may do not need to calculate it for the second time. act_baseline, _, logp, _ = self.get_action(obs, deterministic=True) # advantage _, rew_baseline, _, _ = self.env.step(act_baseline, obs=obs) advantage = rew.to(device).float() - rew_baseline.to(device).float() loss = logp * Variable(advantage).expand_as(act) loss = loss.mean() return loss def update(self, batch_obs, batch_acts, batch_rews): """take a single policy gradient update step for a batch""" self.optimizer.zero_grad() batch_loss = self.compute_loss( obs=torch.as_tensor(batch_obs, dtype=torch.float32), act=torch.as_tensor(batch_acts, dtype=torch.int32), rew=torch.as_tensor(batch_rews, dtype=torch.int32), ) batch_loss.backward() self.optimizer.step() return batch_loss
if convert_to_2d: inp_set = get_2d_converted_data(inp_set) inp_set = torch.from_numpy(inp_set).float() file_name = os.path.basename(inp_file) out_file = os.path.join(out_dir, file_name) data.append((inp_set, out_file)) return data def save_pred(model, data): model.eval() for image, file_path in data: img = image.cuda(cuda) pred = model(img) pred = pred.detach().cpu().numpy()[0] pred = (pred * 255) .astype(np.uint8) # save_path = file_path.replace('.mat', '.png') # cv2.imwrite(save_path, pred) pred = pred.transpose((1, 2, 0)) savemat(file_path, {'crop_g': pred}) if __name__ == '__main__': cuda = torch.device('cuda') model = UNet(n_channels=45, n_classes=3) print("{} Parameters in total".format(sum(x.numel() for x in model.parameters()))) model.cuda(cuda) model.load_state_dict(torch.load(model_loc+"Model_Final_999_3_5.pkl")) model.eval() model.cuda(cuda) data = get_images() save_pred(model, data)
class UNetObjPrior(nn.Module): """ Wrapper around UNet that takes object priors (gaussians) and images as input. """ def __init__(self, params, depth=5): super(UNetObjPrior, self).__init__() self.in_channels = 4 self.model = UNet(1, self.in_channels, depth, cuda=params['cuda']) self.params = params self.device = torch.device('cuda' if params['cuda'] else 'cpu') def forward(self, im, obj_prior): x = torch.cat((im, obj_prior), dim=1) return self.model(x) def train(self, dataloader_train, dataloader_val): since = time.time() best_loss = float("inf") dataloader_train.mode = 'train' dataloader_val.mode = 'val' dataloaders = {'train': dataloader_train, 'val': dataloader_val} optimizer = optim.SGD(self.model.parameters(), momentum=self.params['momentum'], lr=self.params['lr'], weight_decay=self.params['weight_decay']) train_logger = LossLogger('train', self.params['batch_size'], len(dataloader_train), self.params['out_dir']) val_logger = LossLogger('val', self.params['batch_size'], len(dataloader_val), self.params['out_dir']) loggers = {'train': train_logger, 'val': val_logger} # self.criterion = WeightedMSE(dataloader_train.get_classes_weights(), # cuda=self.params['cuda']) self.criterion = nn.MSELoss() for epoch in range(self.params['num_epochs']): print('Epoch {}/{}'.format(epoch, self.params['num_epochs'] - 1)) print('-' * 10) # Each epoch has a training and validation phase for phase in ['train', 'val']: if phase == 'train': #scheduler.step() self.model.train() else: self.model.eval() # Set model to evaluate mode running_loss = 0.0 running_corrects = 0 # Iterate over data. samp = 1 for i, data in enumerate(dataloaders[phase]): # zero the parameter gradients optimizer.zero_grad() # forward # track history if only in train with torch.set_grad_enabled(phase == 'train'): out = self.forward(data.image, data.obj_prior) loss = self.criterion(out, data.truth) # backward + optimize only if in training phase if phase == 'train': loss.backward() optimizer.step() loggers[phase].update(epoch, samp, loss.item()) samp += 1 loggers[phase].print_epoch(epoch) # Generate train prediction for check if phase == 'train': path = os.path.join(self.params['out_dir'], 'previews', 'epoch_{:04d}.jpg'.format(epoch)) data = dataloaders['val'].sample_uniform() pred = self.forward(data.image, data.obj_prior) im_ = data.image[0] truth_ = data.truth[0] pred_ = pred[0, ...] utls.save_tensors(im_, pred_, truth_, path) if phase == 'val' and (loggers['val'].get_loss(epoch) < best_loss): best_loss = loggers['val'].get_loss(epoch) loggers[phase].save('log_{}.csv'.format(phase)) # save checkpoint if phase == 'val': is_best = loggers['val'].get_loss(epoch) <= best_loss path = os.path.join(self.params['out_dir'], 'checkpoint.pth.tar') utls.save_checkpoint( { 'epoch': epoch + 1, 'state_dict': self.model.state_dict(), 'best_loss': best_loss, 'optimizer': optimizer.state_dict() }, is_best, path=path) def load_checkpoint(self, path, device='gpu'): if (device != 'gpu'): checkpoint = torch.load(path, map_location=lambda storage, loc: storage) else: checkpoint = torch.load(path) self.model.load_state_dict(checkpoint['state_dict'])
LE_512 = cropImage(LE_img, IMG_SHAPE[0], IMG_SHAPE[1]) sample_le = {} for le_512 in LE_512: tiles = crop_prepare(le_512, CROP_STEP, IMG_SIZE) for n, img in enumerate(tiles): if n not in sample_le: sample_le[n] = [] img = transform.resize(img, (IMG_SIZE * 2, IMG_SIZE * 2), preserve_range=True, order=3) sample_le[n].append(img) SC_UNET = UNet(n_channels=15, n_classes=1) print("{} paramerters in total".format( sum(x.numel() for x in SC_UNET.parameters()))) SC_UNET.cuda(cuda) SC_UNET.load_state_dict(torch.load(model_path)) # SC_UNET.load_state_dict(torch.load(os.path.join(dir_path,"model","HE_HER_mito","HE_X2_HER_0825.pkl"))) SC_UNET.eval() SRRFDATASET = ReconsDataset(img_dict=sample_he, transform=ToTensor(), in_norm=LE_in_norm, img_type=".tif", in_size=256) test_dataloader = torch.utils.data.DataLoader( SRRFDATASET, batch_size=1, shuffle=False, pin_memory=True) # better than for loop result = np.zeros((256, 256, len(SRRFDATASET))) for batch_idx, items in enumerate(test_dataloader):
class Train(object): def __init__(self, configs): self.batch_size = configs.get("batch_size", "16") self.epochs = configs.get("epochs", "100") self.lr = configs.get("lr", "0.0001") device_args = configs.get("device", "cuda") self.device = torch.device( "cpu" if not torch.cuda.is_available() else device_args) self.workers = configs.get("workers", "4") self.vis_images = configs.get("vis_images", "200") self.vis_freq = configs.get("vis_freq", "10") self.weights = configs.get("weights", "./weights") if not os.path.exists(self.weights): os.mkdir(self.weights) self.logs = configs.get("logs", "./logs") if not os.path.exists(self.weights): os.mkdir(self.weights) self.images_path = configs.get("images_path", "./data") self.is_resize = config.get("is_resize", False) self.image_short_side = config.get("image_short_side", 256) self.is_padding = config.get("is_padding", False) is_multi_gpu = config.get("DateParallel", False) pre_train = config.get("pre_train", False) model_path = config.get("model_path", './weights/unet_idcard_adam.pth') # self.image_size = configs.get("image_size", "256") # self.aug_scale = configs.get("aug_scale", "0.05") # self.aug_angle = configs.get("aug_angle", "15") self.step = 0 self.dsc_loss = DiceLoss() self.model = UNet(in_channels=Dataset.in_channels, out_channels=Dataset.out_channels) if pre_train: self.model.load_state_dict(torch.load(model_path, map_location=self.device), strict=False) if is_multi_gpu: self.model = nn.DataParallel(self.model) self.model.to(self.device) self.best_validation_dsc = 0.0 self.loader_train, self.loader_valid = self.data_loaders() self.params = [p for p in self.model.parameters() if p.requires_grad] self.optimizer = optim.Adam(self.params, lr=self.lr, weight_decay=0.0005) # self.optimizer = torch.optim.SGD(self.params, lr=self.lr, momentum=0.9, weight_decay=0.0005) self.scheduler = lr_scheduler.LR_Scheduler_Head( 'poly', self.lr, self.epochs, len(self.loader_train)) def datasets(self): train_datasets = Dataset( images_dir=self.images_path, # image_size=self.image_size, subset="train", # train transform=get_transforms(train=True), is_resize=self.is_resize, image_short_side=self.image_short_side, is_padding=self.is_padding) # valid_datasets = train_datasets valid_datasets = Dataset( images_dir=self.images_path, # image_size=self.image_size, subset="validation", # validation transform=get_transforms(train=False), is_resize=self.is_resize, image_short_side=self.image_short_side, is_padding=False) return train_datasets, valid_datasets def data_loaders(self): dataset_train, dataset_valid = self.datasets() loader_train = DataLoader( dataset_train, batch_size=self.batch_size, shuffle=True, drop_last=True, num_workers=self.workers, ) loader_valid = DataLoader( dataset_valid, batch_size=1, drop_last=False, num_workers=self.workers, ) return loader_train, loader_valid @staticmethod def dsc_per_volume(validation_pred, validation_true): assert len(validation_pred) == len(validation_true) dsc_list = [] for p in range(len(validation_pred)): y_pred = np.array([validation_pred[p]]) y_true = np.array([validation_true[p]]) dsc_list.append(dsc(y_pred, y_true)) return dsc_list @staticmethod def get_logger(filename, verbosity=1, name=None): level_dict = {0: logging.DEBUG, 1: logging.INFO, 2: logging.WARNING} formatter = logging.Formatter( "[%(asctime)s][%(filename)s][line:%(lineno)d][%(levelname)s] %(message)s" ) logger = logging.getLogger(name) logger.setLevel(level_dict[verbosity]) fh = logging.FileHandler(filename, "w") fh.setFormatter(formatter) logger.addHandler(fh) sh = logging.StreamHandler() sh.setFormatter(formatter) logger.addHandler(sh) return logger def train_one_epoch(self, epoch): self.model.train() loss_train = [] for i, data in enumerate(self.loader_train): self.scheduler(self.optimizer, i, epoch, self.best_validation_dsc) x, y_true = data x, y_true = x.to(self.device), y_true.to(self.device) y_pred = self.model(x) # print('1111', y_pred.size()) # print('2222', y_true.size()) loss = self.dsc_loss(y_pred, y_true) loss_train.append(loss.item()) self.optimizer.zero_grad() loss.backward() self.optimizer.step() # lr_scheduler.step() if self.step % 200 == 0: print('Epoch:[{}/{}]\t iter:[{}]\t loss={:.5f}\t '.format( epoch, self.epochs, i, loss)) self.step += 1 def eval_model(self, patience): self.model.eval() loss_valid = [] validation_pred = [] validation_true = [] # early_stopping = EarlyStopping(patience=patience, verbose=True) for i, data in enumerate(self.loader_valid): x, y_true = data x, y_true = x.to(self.device), y_true.to(self.device) # print(x.size()) # print(333,x[0][2]) with torch.no_grad(): y_pred = self.model(x) loss = self.dsc_loss(y_pred, y_true) # print(y_pred.shape) mask = y_pred > 0.5 mask = mask * 255 mask = mask.cpu().numpy()[0][0] # print(mask) # print(mask.shape()) cv2.imwrite('result.png', mask) loss_valid.append(loss.item()) y_pred_np = y_pred.detach().cpu().numpy() validation_pred.extend( [y_pred_np[s] for s in range(y_pred_np.shape[0])]) y_true_np = y_true.detach().cpu().numpy() validation_true.extend( [y_true_np[s] for s in range(y_true_np.shape[0])]) # early_stopping(loss_valid, self.model) # if early_stopping.early_stop: # print('Early stopping') # import sys # sys.exit(1) mean_dsc = np.mean( self.dsc_per_volume( validation_pred, validation_true, )) # print('mean_dsc:', mean_dsc) if mean_dsc > self.best_validation_dsc: self.best_validation_dsc = mean_dsc torch.save(self.model.state_dict(), os.path.join(self.weights, "unet_xia_adam.pth")) print("Best validation mean DSC: {:4f}".format( self.best_validation_dsc)) def main(self): # print('train is begin.....') # print('load data end.....') # loaders = {"train": loader_train, "valid": loader_valid} for epoch in tqdm(range(self.epochs), total=self.epochs): self.train_one_epoch(epoch) self.eval_model(patience=10) torch.save(self.model.state_dict(), os.path.join(self.weights, "unet_final.pth"))
SRRFDATASET2 = ReconsDataset( train_in_path= "/media/star/LuhongJin/UNC_data/SIM/ALL_data/microtubule/Training_Testing/testing_HE_X2/", train_gt_path= "/media/star/LuhongJin/UNC_data/SIM/ALL_data/microtubule/Training_Testing/testing_HER/", transform=ToTensor(), img_type='tif', in_size=256) validation_dataloader = torch.utils.data.DataLoader( SRRFDATASET2, batch_size=batch_size, shuffle=True, pin_memory=True) # better than for loop model = UNet(n_channels=15, n_classes=1) print("{} paramerters in total".format( sum(x.numel() for x in model.parameters()))) model.cuda(cuda) optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate, betas=(0.9, 0.999)) loss_all = np.zeros((2000, 4)) for epoch in range(2000): mae_m, mae_s = val_during_training(train_dataloader) loss_all[epoch, 0] = mae_m loss_all[epoch, 1] = mae_s mae_m, mae_s = val_during_training(validation_dataloader) loss_all[epoch, 2] = mae_m loss_all[epoch, 3] = mae_s
#train_loader = torch.utils.data.DataLoader(train_data, batch_size=batch_size, shuffle=True, pin_memory=False) # better than for loop #val_loader = torch.utils.data.DataLoader(test_data, batch_size=batch_size, shuffle=False, pin_memory=False) # better than for loop #torch.autograd.set_detect_anomaly(True) if True: # __name__ == '__main__': #opt = parser.parse_args() #train_set = TrainDatasetFromFolder('data/DIV2K_train_HR', crop_size=CROP_SIZE, upscale_factor=UPSCALE_FACTOR) #val_set = ValDatasetFromFolder('data/DIV2K_valid_HR', upscale_factor=UPSCALE_FACTOR) #train_loader = DataLoader(dataset=train_set, num_workers=4, batch_size=64, shuffle=True) #val_loader = DataLoader(dataset=val_set, num_workers=4, batch_size=1, shuffle=False) netG = UNet(n_channels=15, n_classes=1) #print(summary(netG,(15,128,128))) print('# generator parameters:', sum(param.numel() for param in netG.parameters())) netD = Discriminator() print('# discriminator parameters:', sum(param.numel() for param in netD.parameters())) #print(summary(netD,(1,256,256))) generator_criterion = GeneratorLoss() if torch.cuda.is_available(): netG.cuda() netD.cuda() generator_criterion.cuda() optimizerG = optim.Adam(netG.parameters()) optimizerD = optim.Adam(netD.parameters())
nn.init.constant_(m.bias.data, 0) net = UNet(3, 1).to(device) # Apply the weights_init function to randomly initialize all weights net.apply(weights_init) print(net) print('parameters:', get_layer_param(net)) # Initialize BCELoss function criterion = nn.MSELoss() # Setup Adam optimizers for both G and D # optimizer = optim.Adam(net.parameters(), lr=lr) optimizer = optim.Adam(net.parameters(), lr=lr, betas=(0.5, 0.999)) # optimizer = optim.SGD(net.parameters(), lr=lr, momentum=0.9, weight_decay=0.0005) # define a method to save loss image def save_loss_image(list_loss, path): plt.figure(figsize=(10, 5)) plt.title("Generator and Discriminator Loss During Training") plt.plot(list_loss, label="loss") plt.xlabel("iterations") plt.ylabel("Loss") plt.legend() plt.savefig(os.path.join(path, 'loss.jpg')) plt.close() list_loss = []
# Load model if available if(resume==True): print('Resuming training....') generator.load_state_dict(torch.load(os.path.join(model_path,'model_gen_latest'))) discriminator_g.load_state_dict(torch.load(os.path.join(model_path,'model_gdis_latest'))) discriminator_l.load_state_dict(torch.load(os.path.join(model_path,'model_ldis_latest'))) device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") generator = generator.to(device) discriminator_g = discriminator_g.to(device) discriminator_l = discriminator_l.to(device) optimizer_g = optim.Adam(discriminator_g.parameters(), lr=0.00005) optimizer_l = optim.Adam(discriminator_l.parameters(), lr=0.00005) gen_optimizer = optim.Adam(generator.parameters(), lr=0.0002) lossdis = nn.BCELoss() lossgen = FocalLoss() lamda = 75 data_loader = load_images(data_path) num_epochs = 2000 for epoch in range(num_epochs): print() for n_batch, (real_data, gt_data) in enumerate(data_loader): # 1. Train Discriminator N = real_data.size(0)
return model if __name__ == '__main__': lr = 0.001 model = UNet(n_channels=1) #num_ftrs = model.fc.in_features # Here the size of each output sample is set to 2 # Alternatively it can be generalized to nn.Linear(num_ftrs, len(class_names)) #model.fc = nn.Linear(num_ftrs, 2) model = model.to(device) criterion = nn.CrossEntropyLoss() # Observe that all parameters are being optimized optimizer_ft = optim.Adam(model.parameters(), lr=lr) #, momentum=0.9) # Decay LR by a factor of 0.1 every 7 epochs exp_lr_scheduler = lr_scheduler.StepLR(optimizer_ft, step_size=7, gamma=0.1) # Finetune training the convnet and evaluation model = train_model(model, criterion, optimizer_ft, exp_lr_scheduler, num_epochs=25)
writer = SummaryWriter() image = Image.open('./ht2-c2.jpg') out = TF.to_tensor(image) out = out.reshape(1, 3, 640, 640) inp = torch.rand(1, 3, 640, 640) fig = plt.figure() plt.imshow(out[0].permute(1, 2, 0).numpy()) # plt.show writer.add_figure("Ground Truth", fig) fig = plt.figure() plt.imshow(inp[0].permute(1, 2, 0).numpy()) writer.add_figure("Input", fig) num_iter = 500 writer.add_scalar("Number_of_Iterations", num_iter) model = UNet(3, 3) if torch.cuda.is_available(): model.cuda() criterion = nn.MSELoss() learning_rate = 0.1 writer.add_scalar("Learning_Rate", learning_rate) optimizer = torch.optim.SGD(model.parameters(), lr=learning_rate) train(num_iter, inp, out, model, optimizer, criterion) writer.close()
num_workers=1, pin_memory=True) # logging training overview print('-----\n Start training:') print( f'epochs: {args.epochs} \t batch size: {args.batch_size} \t learning rate: {args.learning_rate} \t' ) print( f'training size: {n_train} \t validation size: {n_val} \t checkpoints_dir: {args.checkpoints_dir} \t images downscale: {args.down_scale}' ) print('-----') ## --- Set up training global_step = 0 optimizer = optim.RMSprop(net.parameters(), lr=args.learning_rate, weight_decay=1e-8) if net.n_classes > 1: criterion = nn.CrossEntropyLoss() else: criterion = nn.BCEWithLogitsLoss() ## --- Start training epoch_loss_list = [] val_score_list = [] num_batches_per_epoch = len(dataset) // args.batch_size for epoch in range(args.epochs): net.train() epoch_loss = 0