search

PaddlePaddle / models

Officially maintained, supported by PaddlePaddle, including CV, NLP, Speech, Rec, TS, big models and so on.
Apache License 2.0
6.91k stars 2.91k forks source link

windows下CPU运行RCNN的infer.py出错 #2821

Closed hong3731 closed 5 years ago

hong3731 commented 5 years ago

在windows下CPU运行RCNN的infer.py出错,下载的权重是按照pretrained里面的download.sh的地址下载的,使用的paddle 1.5版本,报错的信息如下: C++ Callstacks: holder_ should not be null Tensor not initialized yet when Tensor::type() is called. at [E:\release_cuda87\paddle\paddle/fluid/framework/tensor.h:139] PaddlePaddle Call Stacks: Windows not support stack backtrace yet. 配置里面已经把设置了使用CPU add_arg('use_gpu', bool, False, "Whether use GPU.") 请问这个是什么问题呢?这个不支持windows的?

FDInSky commented 5 years ago

这是在PaddleDetection下出的错误吗？

hong3731 commented 5 years ago

不是，运行的是models /PaddleCV/rcnn/infer.py

FDInSky commented 5 years ago

经过测试，可以正常运行；根据你的报错信息，请检查下预训练权重的加载路径信息；

hong3731 commented 5 years ago

感谢回复，我检查了一下模型路径，代码中有以下的路径检查的吧，运行中没有报这个错误，是不是可以说明模型路径没有问题的呢？ exe = fluid.Executor(place)

yapf: disable

if not os.path.exists(cfg.pretrained_model):
    raise ValueError("Model path [%s] does not exist." % (cfg.pretrained_model))
JiaXiao243 commented 5 years ago

请问您具体的python版本是？我这边CPU跑infer，运行python infer.py --dataset=coco2017 --pretrained_model=model_iter179999 --image_path=dataset/coco/val2017/000000000139.jpg --draw_threshold=0.6 --MASK_ON=true --use_gpu false，运行结果如下： image

hong3731 commented 5 years ago

你好,pretrained_model是要自己训练的么?预训练的不行不能直接使用? 我下载了三个模型: 1.wget http://paddlemodels.bj.bcebos.com/faster_rcnn/imagenet_resnet50_fusebn.tar.gz 2.https://paddlemodels.bj.bcebos.com/object_detection/mask_rcnn_r50_1x.tar 3.https://paddlemodels.bj.bcebos.com/object_detection/mask_rcnn_se154_vd_fpn_s1x.tar

第一个预训练的权重报 C++ Callstacks: holder_ should not be null Tensor not initialized yet when Tensor::type() is called. at [E:\release_cuda87\paddle\paddle/fluid/framework/tensor.h:139] PaddlePaddle Call Stacks: Windows not support stack backtrace yet.

第二个预训练可以运行,但是结果貌似是错误的 image

第三个预训也是报和第一个一样的错误 infer() C++ Callstacks: holder_ should not be null Tensor not initialized yet when Tensor::type() is called. at [E:\release_cuda87\paddle\paddle/fluid/framework/tensor.h:139] PaddlePaddle Call Stacks: Windows not support stack backtrace yet. 请教一下模型在那里下载的呢?