My YOLOv8-based AI-NVR only detects person

Accelerated Computing Intelligent Video Analytics Metropolis Microservices for Jetson
, , ,
1

• Hardware Platform (Jetson Orin NX16)
• DeepStream Version 7.0
• JetPack Version 6.0 [L4T 36.3.0]

I deployed an AI-NVR based on the official YOLOv8s model. Everything works smoothly, but for some reason, I can only get information about the “Person” class. When using ROI in VST’s web UI, it doesn’t respond to objects other than people. Looking at Redis, I can also only see detection information for the “Person” class.

Besides setting num-detected-classes=80 and having 80 categories in label.txt, are there any other parameters that might affect object classes?

image1
image1507×317 51.7 KB

image2
image2438×382 44.2 KB

2

Can you share the log of: docker logs deepstream? I want to conform YOLOv8s model is used.

1 Like
3

Thanks for your reply! This is my log.
deepstream-json.log (4.3 MB)

5

Below is Redis metadata in my side with YOLOv8s on JPS 2.0. Please follow this guide to have a try with YOLOv8s DLA on JPS 2.0 with JetPack 6.1 or JetPack 6.2: DeepStream Perception — Jetson Platform Services documentation

     2) 1) "message"
        2) "{\n  \"version\" : \"4.0\",\n  \"id\" : \"3678\",\n  \"@timestamp\" : \"2025-01-20T03:20:50.779Z\",\n  \"sensorId\" : \"11\",\n  \"objects\" : [\n    \"608|81.8097|483.152|267.326|893.976|person|#|||||||0\",\n    \"610|0.337343|486.304|138.885|935.49|person|#|||||||0\",\n    \"579|892.617|494.06|1918.5|1069.31|car|#|||||||0\",\n    \"598|603.155|490.543|691.194|551.978|car|#|||||||0\",\n    \"609|716.271|482.27|807.643|539.226|car|#|||||||0\",\n    \"611|1647.82|485.248|1838.59|551.355|car|#|||||||0\",\n    \"616|845.148|486.746|979.592|585.681|car|#|||||||0\",\n    \"613|574.822|483.295|621.895|518.565|car|#|||||||0\",\n    \"614|537.576|477.409|576.961|509.638|car|#|||||||0\"\n  ]\n}"
226) 1) "1737343250852-0"
     2) 1) "message"
        2) "{\n  \"version\" : \"4.0\",\n  \"id\" : \"3679\",\n  \"@timestamp\" : \"2025-01-20T03:20:50.813Z\",\n  \"sensorId\" : \"11\",\n  \"objects\" : [\n    \"609|716.406|482.241|807.779|539.198|car|#|||||||0\",\n    \"614|537.906|477.765|577.292|509.994|car|#|||||||0\",\n    \"610|9.55377|483.659|148.102|932.845|person|#|||||||0\",\n    \"611|1640.83|485.102|1831.59|551.209|car|#|||||||0\",\n    \"608|86.2338|483.299|271.75|894.123|person|#|||||||0\",\n    \"613|572.909|482.561|619.982|517.831|car|#|||||||0\",\n    \"598|603.098|490.934|691.137|552.369|car|#|||||||0\",\n    \"616|844.786|487.454|979.231|586.389|car|#|||||||0\"\n  ]\n}"
227) 1) "1737343250890-0"
     2) 1) "message"
        2) "{\n  \"version\" : \"4.0\",\n  \"id\" : \"3680\",\n  \"@timestamp\" : \"2025-01-20T03:20:50.846Z\",\n  \"sensorId\" : \"11\",\n  \"objects\" : [\n    \"608|89.4462|482.995|267.235|882.332|person|#|||||||0\",\n    \"610|0|488.834|134.059|928.61|person|#|||||||0\",\n    \"579|940.481|505.923|1914.98|1069.4|car|#|||||||0\",\n    \"598|606.112|492.348|695.692|554.575|car|#|||||||0\",\n    \"616|861.089|489.377|1033.23|605.134|car|#|||||||0\",\n    \"611|1623.92|485.238|1814.43|549.784|car|#|||||||0\",\n    \"609|721.363|482.979|812.86|541.435|car|#|||||||0\",\n    \"613|574.61|482.25|620.762|518.229|car|#|||||||0\",\n    \"614|538.125|474.836|577.136|507.814|car|#|||||||0\"\n  ]\n}"
1 Like