Re: [FFmpeg-devel] [PATCH 4/8] libavfilter/dnn: determine dnn output during execute_model instead of set_input_output

2019-04-16 Thread Guo, Yejun
> -Original Message- > From: Steven Liu [mailto:l...@chinaffmpeg.org] > Sent: Tuesday, April 16, 2019 8:04 PM > To: FFmpeg development discussions and patches > Cc: Steven Liu ; Guo, Yejun > Subject: Re: [FFmpeg-devel] [PATCH 4/8] libavfilter/dnn: determine d

Re: [FFmpeg-devel] [PATCH 4/8] libavfilter/dnn: determine dnn output during execute_model instead of set_input_output

2019-04-16 Thread Steven Liu
> 在 2019年4月2日,22:29,Guo, Yejun 写道: > > Currently, within interface set_input_output, the dims/memory of the > tensorflow > dnn model output is determined by executing the model with zero input, > actually, the output dims might vary with different input data for networks > such as object

Re: [FFmpeg-devel] [PATCH 4/8] libavfilter/dnn: determine dnn output during execute_model instead of set_input_output

2019-04-16 Thread Steven Liu
> 在 2019年4月2日,22:29,Guo, Yejun 写道: > > Currently, within interface set_input_output, the dims/memory of the > tensorflow > dnn model output is determined by executing the model with zero input, > actually, the output dims might vary with different input data for networks > such as object

[FFmpeg-devel] [PATCH 4/8] libavfilter/dnn: determine dnn output during execute_model instead of set_input_output

2019-04-02 Thread Guo, Yejun
Currently, within interface set_input_output, the dims/memory of the tensorflow dnn model output is determined by executing the model with zero input, actually, the output dims might vary with different input data for networks such as object detection models faster-rcnn, ssd and yolo. This patch