Mr.Felix

mxnet.base.MXNetError: [16:31:18] src/operator/nn/./cudnn/cudnn_convolution-inl.h:449: Check failed: e == CUDNN_STATUS_SUCCESS (3 vs. 0) cuDNN: CUDNN_STATUS_BAD_PARAM

1

I make the mxnet and when I runned the FCIS,I got the following problem,and I suppose it's because cudnn version.cuda8.0 cudnn6 loading annotations into memory... Done (t=0.52s) creating index... index...

Is the stride or feature map size influence the NME?

2

I trained on my dataset and use the input size 64X80, and found the NME is larger than the regression method before used, and when I modified the input size...

Use TAO can I modify the model structure by myself?

1

I have train yolov4 example successfully and thank you for your great work! But if the pretrained model you provided is not meet the need of my project,can I modify...

@dji-dev 使用机场3连接同一网段的服务器（边缘设备）并完成绑定，但在拉流测试过程发给发现绑定状态会频繁切换导致拉流卡死或失败。错误显示码流解密失败，经过排查是会话密钥频繁变更引起，而会话密钥频繁变更是绑定状态频繁变化导致。但是机场和边缘设备在同一局域网，网络稳定性应该没问题。想问下是否是机场和ESDK版本不匹配或是一些第三方库（libssh）不兼容导致？码流目前能正常拉画面也是正常，但就是会频繁卡顿和闪退。  日志： [h264 @ 0x5bf72f150090] SEI type 101 size 792 truncated at 768 [h264 @ 0x5bf72f150090] SEI type 101 size 792 truncated at 752 Image sent to display...