Chaotic predictions with deepstream-app

Hello, I recently bought a Jetson Xavier and I installed Jetpack 4.1.1 and DeepStream 3.0 on it.

Im trying to run predictions with yolo3 on a video stream (Im using the sample_720 that comes with DeepStream) using the objectDetector_YoloV3. But instead of getting the correct predictions, Im getting a lot of random predictions in the top left corner of the video. See this link for a screenshot of the predictions:

Im not getting any strange warnings or error in the terminal, and Im using trt engine generated with the trt-yolo-app for the vanilla yolov3 that comes with DeepStream. The predictions on images that I get using that engine are correct, so I dont think the problem is there.

The only warnings I get are this:

WARNING from sink_sub_bin_sink1: A lot of buffers are being dropped.
Debug info: gstbasesink.c(2902): gst_base_sink_is_too_late (): /GstPipeline:pipeline/GstBin:processing_bin_0/GstBin:sink_bin/GstBin:sink_sub_bin1/GstEglGlesSink:sink_sub_bin_sink1:
There may be a timestamping problem, or this computer is too slow.

But those may be appearing because of the insane amount of predictions that can be see on the previous link.

Im also using the deepstream_app_config_yoloV3.txt thats on the current master in github.

Somebody knows what could be wrong?

Hi jpdevicente555,

Please refer to below topics to see if can help:

Besides, we’re going to have a new DeepStream SDK release, targeting at the late June or early July, please stay tuned.


Ok thanks, I’ll leave the Jetson on the side and come back to it once the new SDK its released.