Hi Team,
We are trying to run the Deep Learning Inference using Gstreamer SNPE plugin(i.e. qtimlesnpe) from terminal using gst-launch tool.
We have tried running command given below,
gst-launch-1.0 v4l2src ! jpegdec ! videoconvert ! qtimlesnpe config=/home/rb5/mle_snpe.config model=/home/rb5/mobilenet_v1.dlc labels=/home/rb5/labelmap.txt postprocessing=segmentation ! videoconvert ! jpegenc ! filesink location=image.jpeg
After running command given above, its causing the below warning, and terminating the execution:
WARNING: erroneous pipeline: no element "qtimlesnpe"
Can you please suggest any solution on this?
Hi,
Thanks for reaching out.
According to the newest created image QRB5165.1.0.0-201116, qtimlesnpe elelemnt is not included yet. There will be a set of C/C++ gstreamer SNPE sample programs released soon. qtimlesnpe element should be included in the newer image when the sample code released. At this time, gstreamer SNPE plugin is not available.
Please use 'gst-inspect-1.0 | grep qti' to check the installed qti elements in the newer image.
Best regards,
Stan
Hello,
Could you please let us know by when do you expect to release the new image with "qtimlesnpe" element included in it? We are working with one of the deep learning segmentation models and we really need this gstreamer element to load the dlc model and run inference on the camera stream. We have no idea that this element is not included in the current image and it is not mentioned in the documentation. It will be really helpful if you could give us some estimation by when the new image will be released. Thank you.
Regards
Rajesh
Hi
Can you please suggest command with qtimlesnpe plug-in without need of live streaming from camera. If have pre-built .dlc file, possible to use it and validate this plug-in on RB5 platform without connecting camera and display?
Hi,
Could you suggest the gstreamer pipeline using qtimlesnpe plugin for the prebuilt model in .dlc file.I have the pipeline that has been stuck with prerolling state...What is the reason for that?