1. 数据集预处理
思路是仿照一个结构和Synapse相似的数据集。
如果我的数据集是这样的
D:/compare/US30K0817/BUSIS/
├── images/ # 原始图像 (PNG格式)
│ ├── case0001.png
│ ├── case0002.png
│ └── ...
└── labels/ # 分割标签 (PNG格式)
├── case0001.png
├── case0002.png
└── ...运行preprocess/data_process.py
修改三个参数
parser.add_argument('--src_path', type=str, default='D:/compare/US30K0817/BUSIS',
help='Path to the root directory containing "images" and "labels" folders')
parser.add_argument('--dst_path', type=str, default='D:/compare/0816/train_npz_BUSI2',
help='Output directory for processed files')
parser.add_argument('--use_normalize', action='store_true', default=True,
help='Normalize image pixel values to [0,1]')在dst_path输出路径下有三个文件夹
D:\COMPARE\0816\TRAIN_NPZ_BUSI2
├─lists
│ └─lists_custom
├─test_vol_h5
└─train_npztrain_npz中输出的.npz文件是二维的，每个文件对应一个二维切片，是train.py中的root_path参数。test_vol_h5中验证集的.h5文件是三维的，保存了完整的3D体积数据，没有用到。lists\lists_custom**中有train.txttest.txt**，需要将整个lists_custom文件夹复制到项目文件夹D:\compare\H-SAM-main\lists中。
2. 调整参数并运行
CUDA_VISIBLE_DEVICES="0" python train.py \
--root_path "自定义数据集/train_npz" \
--output "./results/custom/" \
--split "train" \
--batch_size 8 \
--base_lr 0.0026 \
--img_size 512 \
--warmup \
--AdamW \
--max_epochs 300 \
--stop_epoch 300 \
--num_classes <您的类别数> \ #二分为2
--dataset "custom" \
--list_dir "./lists/lists_custom" \
--ckpt "checkpoints/sam_vit_b_01ec64.pth"我是手动在train.py文件中修改参数的
parser.add_argument('--root_path', type=str,
default='D:/compare/0816/train_npz_BUSI2/train_npz', help='root dir for data')
parser.add_argument('--output', type=str, default='./results/custom/')
parser.add_argument('--dataset', type=str,
default='Synapse', help='experiment_name')
parser.add_argument('--list_dir', type=str,
default='./lists/lists_custom', help='list dir')
parser.add_argument('--split', type=str,
default='train', help='list dir')
parser.add_argument('--num_classes', type=int,
default=2, help='output channel of network')
parser.add_argument('--max_iterations', type=int,
default=30000, help='maximum epoch number to train')
parser.add_argument('--max_epochs', type=int,
default=30, help='maximum epoch number to train')
parser.add_argument('--stop_epoch', type=int,
default=30, help='maximum epoch number to train')
parser.add_argument('--batch_size', type=int,
default=24, help='batch_size per gpu')
parser.add_argument('--n_gpu', type=int, default=2, help='total gpu')
parser.add_argument('--deterministic', type=int, default=1,
help='whether use deterministic training')
parser.add_argument('--base_lr', type=float, default=0.005,
help='segmentation network learning rate')
parser.add_argument('--img_size', type=int,
default=224, help='input patch size of network input')
parser.add_argument('--seed', type=int,
default=2345, help='random seed')
parser.add_argument('--vit_name', type=str,
default='vit_b', help='select one vit model')
parser.add_argument('--ckpt', type=str, default='checkpoints/sam_vit_b_01ec64.pth',
help='Pretrained checkpoint')
parser.add_argument('--lora_ckpt', type=str, default=None, help='Finetuned lora checkpoint')
parser.add_argument('--rank', type=int, default=5, help='Rank for LoRA adaptation')
parser.add_argument('--warmup', action='store_true', help='If activated, warp up the learning from a lower lr to the base_lr')
parser.add_argument('--warmup_period', type=int, default=250,
help='Warp up iterations, only valid when warmup is activated')
parser.add_argument('--AdamW', action='store_true', help='If activated, use AdamW to finetune SAM model')
parser.add_argument('--module', type=str, default='sam_lora_image_encoder')
parser.add_argument('--dice_param', type=float, default=0.9)