chenjoya / Vehicle_Detection_Recognition

This is a Matlab lesson design for vehicle detection and recognition. Using cifar-10Net to training a RCNN, and finetune AlexNet to classify. Thanks to Cars Dataset:http://ai.stanford.edu/~jkrause/cars/car_dataset.html
202 stars 81 forks source link

还是不懂怎么让它用CPU运行呀 #5

Open ghost opened 7 years ago

ghost commented 7 years ago

不行呀博主,MATLAB提示GPU不支持呀,我可以加下你好友吗?你看看“ There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver is installed.”出错 parallel.internal.gpu.isAnyDeviceSelected

出错 parallel.gpu.GPUDevice.isAvailable (line 108) if parallel.internal.gpu.isAnyDeviceSelected

出错 nnet.internal.cnngpu.isGPUCompatible (line 9) if(canUsePCT() && parallel.gpu.GPUDevice.isAvailable())

出错 trainNetwork (line 46) nnet.internal.cnngpu.isGPUCompatible(true);

原因: The CUDA driver could not be loaded. The library name used was 'nvcuda.dll'. The error was: 找不到指定的模块。

chenjoya commented 7 years ago

Matlab的Deep learning Toolbox不支持CPU训练,但是还是支持网络前向传播的。

ghost commented 7 years ago

那我该怎么办?

WilliamTW commented 7 years ago

Predictor=CarProphet('Model/cifar10NetRCNN.mat','Model/imagenet-caffe-alex.mat','Model/cars_meta.mat'); 把第二个参数(想看看其他分类网络效果)换了以后就一直bug,MATLAB不是特别熟,还望指教

ghost commented 7 years ago

你这是什么意思?还是用的GPU?

chenjoya commented 7 years ago

@WilliamTW 你可以修改Class类里面的东西。因为这两个权重文件我以结构体的形式包装了,所以会读取出错。 @jokezhao 我这边完全可以用CPU运行。我并不清楚你是不是Matlab 2016b

ghost commented 7 years ago

我是2016b呀,我完全按照你博客上的方法运行的http://blog.csdn.net/Mr_Curry/article/details/53160914,然后就遇到了这个问题

ghost commented 7 years ago

然后我不知道怎么改动让它用CPU运行

ghost commented 7 years ago

刚才看了2016b的文档,找不到哪地方支持CPU

ghost commented 7 years ago

大神,你就告诉我吧,我毕设马上就要中期答辩了

ghost commented 7 years ago

QQ:1187597983

chenjoya commented 7 years ago

@jokezhao 我并不清楚如何帮助你,因为这个无关修改为CPU运行... 你如果实在不能运行的话,可以看我的博客上有一篇在Windows下配置Faster-RCNN的文章http://blog.csdn.net/mr_curry/article/details/54745116,你基于微软的Caffe能够很快跑起来这个demo,然后可以自己训练。