gst-inference
gst-inference copied to clipboard
RidgeRun
A GStreamer Deep Learning Inference Framework
In my Mac OSX: ``` meson --version 0.53.0 pkg-config --modversion opencv4 4.3.0 ``` I get the following: ``` ... Found CMake: /usr/local/bin/cmake (3.17.2) Run-time dependency opencv found: NO (tried pkgconfig...
If the backend option is not set correctly, there is no error handling. See for examples: $ gst-launch-1.0 mobilenetv2 name=ynet model-location=/home/mendel/mobilenet_v2_1.0_224/mobilenet_v2_1.0_224.tflite backend=tensorflow backend::input-layer=input backend::output-layer=MobilenetV2/Predictions/Reshape_1 filesrc location=/usr/share/edgetpudemo/video_stream.mp4 ! decodebin ! videoconvert...
Tests for tflite and tensorflow only have a test for negative values but null values aren't being tested
I know this is a dumb and erroneous use case, but we should error out instead of segfaulting. This is an example pipeline: ``` gst-launch-1.0 avfvideosrc ! video/x-raw ! inferencefilter...
I see sporadic segfaults when starting the following pipeline: ``` inferencefilter filter-class=-1 ! inferencedebug name=before ! videoconvert ! tee name=tee tee. ! queue max-size-buffers=3 ! arch.sink_bypass tee. ! queue max-size-buffers=3...
I have a segfault with the following scenario: * R2Iference version: v0.5.0 * GstInference version: v0.7.0 Simple pipe to reproduce the problem: `gst-launch-1.0 tinyyolov2` Backtrace: ``` (lldb) bt * thread...
Hi GstInference team, In the configure process, GstInference searches for OpenCV 2.X or 3.X, in case of systems with OpenCV 4.1 installed, this package is not detected and inference overlay...
In order for Video Inference to emit the "new-inference" signal it needs to have both model and bypass sink pads. We should only need the model pad.
Hi, Is it Resnet50 v1 supported on the plugin yet? I am interested in using that. Also, I have a question about performance measurements. Would anyone know how to measure...
