Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU) Jetson Orin AGX Dev Kit
• DeepStream Version 7.1
• JetPack Version (valid for Jetson only) 6.2.1
• TensorRT Version nvidia-tensorrt 6.2.1+b38 and tensorrt 10.3.0.30-1+cuda12.5
• Issue Type( questions, new requirements, bugs) problem to configure nvdspreprocess in deepstream pipeline
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)
Hi everybody!
I’m building a DeepStream 7.1 pipeline where I have two layers of inference. The PGIE is a yolov8m segmentation and it goes into 2 SGIEs yolov8 classification, each one filter the class_id from the PGIE and infer only on one.
I try to implement a nvdspreprocess to make the 2 SGIEs infer on cropped areas extracted from the full-resolution frame (and not the 832x832 from my first seg model).
My setup:
-
Input video: 3840x2160, demuxed and decoded, then split via
tee. -
Models:
-
nvinfer0— segmentation model, input 832x832. -
nvinfer1— classification model 640x640 “GD_L,” expects inference on croped ROI on full-resolution frames. -
nvinfer2— classification model" “GD_Y” 640x640, same as above.
-
-
Problem: Pipeline fails to start with:
[FATAL] ... nvdspreprocess_start: error: Configuration file parsing failed
Everything seems correct per the documentation — keys, network shapes, tensor names, color formats, etc. I want the nvdspreprocess to run on full-resolution frames before classification models, but the parser refuses the config. We can see some good parsing :
[gst_pipeline_node-5] 0:00:00.279488326 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:823:nvdspreprocess_parse_config_file: Group found property
[gst_pipeline_node-5]
[gst_pipeline_node-5] 0:00:00.279537606 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:266:nvdspreprocess_parse_property_group: Parsed 'target-unique-ids=2' in group 'property'
[gst_pipeline_node-5]
[gst_pipeline_node-5] 0:00:00.279545222 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:266:nvdspreprocess_parse_property_group: Parsed 'target-unique-ids=3' in group 'property'
[gst_pipeline_node-5]
[gst_pipeline_node-5] 0:00:00.279554150 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:300:nvdspreprocess_parse_property_group: Parsed 'network-input-shape=1' in group 'property'
[gst_pipeline_node-5]
[gst_pipeline_node-5] 0:00:00.279558086 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:300:nvdspreprocess_parse_property_group: Parsed 'network-input-shape=3' in group 'property'
[gst_pipeline_node-5]
[gst_pipeline_node-5] 0:00:00.279561606 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:300:nvdspreprocess_parse_property_group: Parsed 'network-input-shape=640' in group 'property'
[gst_pipeline_node-5]
[gst_pipeline_node-5] 0:00:00.279564582 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:300:nvdspreprocess_parse_property_group: Parsed 'network-input-shape=640' in group 'property'
[gst_pipeline_node-5]
[gst_pipeline_node-5] 0:00:00.279571270 34031 0xaaaae0f166a0 INFO nvdspreprocess nvdspreprocess_property_parser.cpp:401:nvdspreprocess_parse_property_group: Parsed tensor-name=input in group 'property'
I give you my logs and configurations here and am avalaible for any additional information :)
error_logs.txt (8.5 KB)
config_infer_classif_GD_L.txt (1.3 KB)
config_preprocess_GD_L.txt (1.5 KB)
config_infer_classif_GD_Y.txt (1.3 KB)
pipeline.txt (2.0 KB)
If i am not clear on a specific point do not hesitate to ask!