萌新,求助
收藏
想做的医学图像灰度图的分割,训练模型的时候出现,.epoch不会增加(我看其他一些博主的代码里都有增加的)?
无论iters取100还是1000结果都一样 ———— Images: 1328 mIoU: 0.5000 Acc: 1.0000 Kappa: nan Dice: 0.5000?
附图:config/yaml文件,和运行记录
batch_size: 4 iters: 100 train_dataset: type: Dataset dataset_root: /home/xyf/MZA/some_model/png train_path: /home/xyf/MZA/some_model/txt/new/train_list.txt num_classes: 2 mode: train transforms: # - type: ResizeStepScaling # min_scale_factor: 0.5 # max_scale_factor: 2.0 # scale_step_size: 0.25 # - type: RandomPaddingCrop # crop_size: [512, 512] - type: RandomHorizontalFlip # - type: RandomDistort # brightness_range: 0.5 # contrast_range: 0.5 # saturation_range: 0.5 - type: Normalize val_dataset: type: Dataset dataset_root: /home/xyf/MZA/some_model/png val_path: /home/xyf/MZA/some_model/txt/new/val_list.txt num_classes: 2 mode: val transforms: - type: Normalize optimizer: type: SGD momentum: 0.9 weight_decay: 4.0e-5 lr_scheduler: type: PolynomialDecay learning_rate: 0.001 end_lr: 0 power: 0.9 loss: types: - type: CrossEntropyLoss coef: [1] model: type: UNet num_classes: 2
2.
------------Environment Information------------- platform: Linux-5.4.0-150-generic-x86_64-with-glibc2.10 Python: 3.8.16 | packaged by conda-forge | (default, Feb 1 2023, 16:01:55) [GCC 11.3.0] Paddle compiled with cuda: True NVCC: Not Available cudnn: 8.2 GPUs used: 1 CUDA_VISIBLE_DEVICES: None GPU: ['GPU 0: NVIDIA GeForce'] GCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0 PaddleSeg: 2.8.0 PaddlePaddle: 2.4.2 OpenCV: 4.5.5 ------------------------------------------------ 2023-06-07 15:57:01 [INFO] ---------------Config Information--------------- batch_size: 4 iters: 100 train_dataset: dataset_root: /home/xyf/MZA/some_model/png mode: train num_classes: 2 train_path: /home/xyf/MZA/some_model/txt/new/train_list.txt transforms: - type: RandomHorizontalFlip - type: Normalize type: Dataset val_dataset: dataset_root: /home/xyf/MZA/some_model/png mode: val num_classes: 2 transforms: - type: Normalize type: Dataset val_path: /home/xyf/MZA/some_model/txt/new/val_list.txt optimizer: momentum: 0.9 type: SGD weight_decay: 4.0e-05 lr_scheduler: end_lr: 0 learning_rate: 0.001 power: 0.9 type: PolynomialDecay loss: coef: - 1 types: - type: CrossEntropyLoss model: num_classes: 2 type: UNet ------------------------------------------------ 2023-06-07 15:57:01 [INFO] Set device: gpu 2023-06-07 15:57:01 [INFO] Use the following config to build model model: num_classes: 2 type: UNet W0607 15:57:01.504024 348892 gpu_resources.cc:61] Please NOTE: device: 0, GPU Compute Capability: 8.6, Driver API Version: 11.3, Runtime API Version: 11.2 W0607 15:57:01.504060 348892 gpu_resources.cc:91] device: 0, cuDNN Version: 8.2. 2023-06-07 15:57:02 [INFO] Use the following config to build train_dataset train_dataset: dataset_root: /home/xyf/MZA/some_model/png mode: train num_classes: 2 train_path: /home/xyf/MZA/some_model/txt/new/train_list.txt transforms: - type: RandomHorizontalFlip - type: Normalize type: Dataset 2023-06-07 15:57:02 [INFO] Use the following config to build val_dataset val_dataset: dataset_root: /home/xyf/MZA/some_model/png mode: val num_classes: 2 transforms: - type: Normalize type: Dataset val_path: /home/xyf/MZA/some_model/txt/new/val_list.txt 2023-06-07 15:57:02 [INFO] If the type is SGD and momentum in optimizer config, the type is changed to Momentum. 2023-06-07 15:57:02 [INFO] Use the following config to build optimizer optimizer: momentum: 0.9 type: Momentum weight_decay: 4.0e-05 2023-06-07 15:57:02 [INFO] Use the following config to build loss loss: coef: - 1 types: - type: CrossEntropyLoss /home/xyf/.conda/envs/paddle/lib/python3.8/site-packages/paddle/nn/layer/norm.py:712: UserWarning: When training, we now always track global mean and variance. warnings.warn( /home/xyf/.conda/envs/paddle/lib/python3.8/site-packages/paddle/fluid/dygraph/math_op_patch.py:275: UserWarning: The dtype of left and right variables are not the same, left dtype is paddle.float32, but right dtype is paddle.int64, the right dtype will convert to paddle.float32 warnings.warn( 2023-06-07 15:57:07 [INFO] [TRAIN] epoch: 1, iter: 10/100, loss: 0.5832, lr: 0.000919, batch_cost: 0.4442, reader_cost: 0.01087, ips: 9.0046 samples/sec | ETA 00:00:39 2023-06-07 15:57:09 [INFO] [TRAIN] epoch: 1, iter: 20/100, loss: 0.0688, lr: 0.000827, batch_cost: 0.2084, reader_cost: 0.02542, ips: 19.1934 samples/sec | ETA 00:00:16 2023-06-07 15:57:11 [INFO] [TRAIN] epoch: 1, iter: 30/100, loss: 0.0184, lr: 0.000735, batch_cost: 0.2071, reader_cost: 0.02405, ips: 19.3181 samples/sec | ETA 00:00:14 2023-06-07 15:57:13 [INFO] [TRAIN] epoch: 1, iter: 40/100, loss: 0.0098, lr: 0.000641, batch_cost: 0.2099, reader_cost: 0.02693, ips: 19.0566 samples/sec | ETA 00:00:12 2023-06-07 15:57:15 [INFO] [TRAIN] epoch: 1, iter: 50/100, loss: 0.0077, lr: 0.000546, batch_cost: 0.2092, reader_cost: 0.02568, ips: 19.1181 samples/sec | ETA 00:00:10 2023-06-07 15:57:15 [INFO] Start evaluating (total_samples: 1328, total_iters: 1328)... 1328/1328 [==============================] - 23s 17ms/step - batch_cost: 0.0169 - reader cost: 9.7707e-05 2023-06-07 15:57:38 [INFO] [EVAL] #Images: 1328 mIoU: 0.5000 Acc: 1.0000 Kappa: 0.0000 Dice: 0.5000 2023-06-07 15:57:38 [INFO] [EVAL] Class IoU: [1. 0.] 2023-06-07 15:57:38 [INFO] [EVAL] Class Precision: [1. 0.] 2023-06-07 15:57:38 [INFO] [EVAL] Class Recall: [1. 0.] 2023-06-07 15:57:38 [INFO] [EVAL] The model with the best validation mIoU (0.5000) was saved at iter 50.
0
收藏
请登录后评论
数据集大小是多少,你可以算出来需要多少个iters是一个epoch
你好请问解决了吗我也遇到了同样问题