I wrote a pipeline using face detector yolov8n model with batch size = 4. I exported the model with --dynamic flag. But, only one stream seems to get the face bboxes in the output, others don't.
Is the current custom parser library written to handle batch processing?