Shaun

I have try it as follows `cd path/to/cudnn` `export CUDNN_DIR=$(pwd)` `export LD_LIBRARY_PATH=${CUDNN_DIR}:$LD_LIBRARY_PATH` `python onnx2tensorrt --configs ****` However, I got the same error

I have delete this two lines in .hpp, but the error is still occur ```python File "/home/cc/Projects/Deploy/mmdeploy/mmdeploy/backend/tensorrt/init_plugins.py", line 33, in load_tensorrt_plugin ctypes.CDLL(lib_path) File "/home/cc/anaconda3/envs/mmdeploy/lib/python3.7/ctypes/__init__.py", line 364, in __init__ self._handle =...

@nvpohanh I got the same error when I build the engine with INT8, static shape is used as the input. 

I got the same problem. I test the speed following your code and provided model, 29.2tasks/s (34ms/img) on single Titan Xp is slower than 24ms/img reported in your paper (RESNET50-FPN...

> @cxiang26 Hi. Thanks for your interest. Which ResNet50 model 6x or 10x? Do you caculate the time about loading image? Thanks for your reply. I use ResNet50 model 6x....

> @cxiang26 Please comment the whole code in the "for" loop. In addtion, you can add time function in https://github.com/JialeCao001/SipMask/blob/master/SipMask-mmdetection/mmdet/models/detectors/single_stage.py#L76. Is inference time only including forward pass? pre- and post-process...

Thanks, I got the same speed 24ms/img as reported in paper.

> @jinfagang and @cxiang26 I have run the following command of SipMask-mmdetection > > python demo/webcam_demo.py configs/sipmask/sipmask_r50_caffe_fpn_ssd_6x.py sipmask_r50_caffe_ssd_6x.pth > > -> I have 720 p resolution video. > -> I...

I have the same question, how to set the initialization of the weight? The training is unstable when we change the hyperparameter of initialization. thanks.

I got the low accuracy as well, which configs could get more precise?