Open yzyMichael opened 3 weeks ago
可以提供下详细的流程吗?
模型产线:面单图像分割进行到“数据准备”阶段的“格式效验”步骤中报报错 模型aistudio地址:https://aistudio.baidu.com/pipeline/step/p-5642d2aa258f 数据集地址:https://aistudio.baidu.com/datasetdetail/299557 数据集在本地CPU机器上使用PaddleX中的 check_dataset 模型效验成功
上传平台的时候是怎么打包的呢?数据集有多大呢?
zip格式包,数据集16.2M
建议打tar包尝试一下哈
Checklist:
描述问题
使用 aistudio 训练自定义数据集时, 报数据格式不正确,失败日志提示:failed to import ujson, using json instead
When using aistudio to train a custom dataset, the data format is incorrect and a failure log prompt appears:“failed to import ujson, using json instead”
复现
c++部署方式
您是否按照文档教程已经正常运行我们提供的demo?
您是否在demo基础上修改代码内容?还请您提供运行的代码
c#部署方式
您使用的模型和数据集是? 使用的模型:PaddleX: instance_segmentation_module 数据集:test_express_dataset alias :express_dataset 数据切分 训练集占比:80%,验证集占比20% trail 80% val:20%
请提供您出现的报错信息及相关log
2024-10-21 10:10:58,300 - pp-pipeline-exec - INFO - u-13199720-pp-p-5642d2aa258f-tt-5ec3c6719ba4 2024-10-21 10:10:58,300 - pp-pipeline-exec - INFO - 执行命令:python /home/aistudio/template_pipelines/general_instance_segmentation_pipeline/instance_segmentation_module/main.py --mode check_dataset --dataset_dir /home/aistudio/data --output_dir /home/aistudio/output --dst_dataset_name express_dataset --model_name Mask-RT-DETR-H --split True --split_train_percent 80 --split_val_percent 20 2024-10-21 10:10:59,548 - pp-pipeline-exec - INFO - ERROR:root:failed to import ujson, using json instead Traceback (most recent call last): File "/home/aistudio/template_pipelines/general_instance_segmentation_pipeline/instance_segmentation_module/base/engine.py", line 31, in wrap func(self, *args, **kwargs) File "/home/aistudio/template_pipelines/general_instance_segmentation_pipeline/instance_segmentation_module/base/engine.py", line 66, in run self.run_dataset_check() File "/home/aistudio/template_pipelines/general_instance_segmentation_pipeline/instance_segmentation_module/base/engine.py", line 108, in run_dataset_check self.dataset_checker() File "/home/aistudio/template_pipelines/general_instance_segmentation_pipeline/instance_segmentation_module/base/dataset_checker/dataset_checker.py", line 29, in call dataset_dir = self.get_dataset_root(self.args.dataset_dir) File "/home/aistudio/template_pipelines/general_instance_segmentation_pipeline/instance_segmentation_module/base/dataset_checker/dataset_checker.py", line 67, in get_dataset_root assert len(dataset_dir) == 1 AssertionError
环境