0

I am working on Nvidia Jetson AGX Xavier within Dockerized container…I want to take input from RTSP stream…it’s encoding type is H264 and .avi video input.The input stream frame size is 1920x1080 (in code I am resizing that into 1280x720)

I have used this GStreamer pipeline

cap = cv2.VideoCapture(‘rtspsrc location=“rtsp_link” latency=200 ! queue ! rtph264depay ! h264parse ! avdec_h264 ! videoconvert ! appsink’, cv2.CAP_GSTREAMER)

It is able to read the frames but after reading frames, frames got completely blurred , that’s why object detection is not happening.

jps
  • 20,041
  • 15
  • 75
  • 79
Tulsee Bisen
  • 1
  • [Under what circumstances may I add "urgent" or other similar phrases to my question, in order to obtain faster answers?](//meta.stackoverflow.com/q/326569) – jps May 04 '22 at 11:31
  • You may better explain your docker container bindings (nvidia runtime, ...) You may have to add caps `application/x-rtp,encoding-name=H264` between rtspsrc and rtph264depay. It may be better to use Jetson HW with `nvv4l2decoder`, then nvvidconv coverting to BGRx or RGBA and finally videoconvert into `video/x-raw, format=BGR` before opencv `... ! appsink drop=1`. – SeB May 11 '22 at 20:16

0 Answers0