"self-collected blood-classification dataset"

"# README## 1. OverviewThis package contains two datasets used in our experiments:1.  **Dataset 1**\\   A self-collected blood-classification dataset pre-processed into a CIFAR-style format.2.  **Dataset 2**\\   A publicly available blood-segmentation dataset.\\   Please refer to its official GitHub repository for detailed   documentation and download instructions.This README describes how to use both datasets.------------------------------------------------------------------------## 2. Dataset 1 (CIFAR-style)### 2.1 StructureThe dataset follows the same data format and file organization asCIFAR-10. It contains blood cell classification images that have been pre-processed into the CIFAR format (224\u00d7224 resolution and 3-channel RGB), derived from our self-collected blood smear samples.\\Directory structure:   cifar-10-batches-py\/   \u2502   \u251c\u2500\u2500 data_batch_1   \u251c\u2500\u2500 data_batch_2   \u251c\u2500\u2500 ...   \u251c\u2500\u2500 test_batch   \u2502   \u2514\u2500\u2500 batches.metaEach batch file is a Python pickle file with fields:-   `data`: image array of shape `(N, 3*224*224)`-   `labels`: class indices### 2.2 Loading InstructionsExample code for loading the dataset:``` pythonimport pickleimport numpy as npfrom torch.utils.data import Datasetload = {}def register_dataset(dataset):   def warpper(f):       load[dataset] = f       return f   return warpper@register_dataset('blood3')def blood3(arg, data_root='.\/data_local\/blood3\/'):   train_transform = transforms.Compose([       transforms.RandomResizedCrop((arg.img_size,arg.img_size),scale=(arg.crop_scale,1)),       CutoutPIL(cutout_factor=0.5),       RandAugment(),       transforms.ToTensor(),   ])   eval_transform = transforms.Compose([       transforms.Resize((arg.img_size,arg.img_size)),       transforms.ToTensor(),   ])   trainset = tv.datasets.CIFAR10(data_root, train=True, transform=train_transform)   evalset = tv.datasets.CIFAR10(data_root, train=False, transform=eval_transform)   return {       'trainset': trainset,       'evalset': evalset   }def create_loaders(arg, trainset, evalset):   print(\"train number:\",len(trainset))   print(\"test number:\",len(evalset))   train_loader = torch.utils.data.DataLoader( trainset,                                               batch_size=arg.train_batch_size,                                               shuffle=True,                                               num_workers=arg.workers,                                               pin_memory=True)   ## test batch loader   eval_loader = torch.utils.data.DataLoader(  evalset,                                               batch_size=arg.eval_batch_size,                                               shuffle=False,                                               num_workers=arg.workers,                                               pin_memory=True,                                               drop_last=False)   print(\"train_loader:\",len(train_loader))   print(\"test_loader:\",len(eval_loader))   return train_loader, eval_loader```Usage example:``` pythondataset = load['blood3'](args, 'data_local\/blood3')train_loader, test_loader = create_loaders(args, dataset['trainset'], dataset['evalset'])```------------------------------------------------------------------------## 3. Dataset 2### 3.1 SourceIt contains three hundred 120\u00d7120 images of WBCs and their color depth is 24 bits. The images were taken by a Motic Moticam Pro 252A optical microscope camera with a N800-D motorized auto-focus microscope, and the blood smears were processed with a newly-developed hematology reagent for rapid WBC staining.\\Please refer to the official repository:**GitHub link:**\\\\[https:\/\/github.com\/zxaoyou\/segmentation_WBC\\]### 3.2 UsageDownload the public dataset following the instructions provided in thelinked repository. Refer to the official dataset documentation for other usage details.------------------------------------------------------------------------ "