Open thenighthunter0 opened 6 years ago
Please make sure that matconvnet had a correct configuration. Errors may be associated with cuda and cudnn.
The input size can be changed according to your design. You can construct a new network based on LapSRN to apply to big size image.
Hi, lemoner20. If I use big size image as input in demo_MSLapSRN, I met this issue as follows.
错误使用 vl_nnconvt vl::nnbias_forward: vl::impl::nnbias_cudnn::forward: cuDNN error [cudnn: 59:"F:/xuexi/SuperResolution-master/SuperResolution-master/LapSRN-matlab/matconvnet/matlab/src/bits/impl/nnbias_cudnn.cu"
(CUDNN_STATUS_NOT_INITIALIZED)]
出错 dagnn.ConvTranspose/forward (line 14)
outputs{1} = vl_nnconvt(...
出错 dagnn.Layer/forwardAdvanced (line 85)
outputs = obj.forward(inputs, {net.params(par).value}) ;
出错 dagnn.DagNN/eval (line 91)
obj.layers(l).block.forwardAdvanced(obj.layers(l)) ;
出错 SR_MSLapSRN (line 52)
net.eval(inputs);
出错 demo_MSLapSRN (line 77)
img_HR = SR_MSLapSRN(img_LR, net, model_scale, test_scale, gpu);
I want to apply big size image as input, how can I solve this issue? A little help will be valued, thank you.