YoloV8 false negatives

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) NVIDIA GeForce RTX 3050 Ti Mobile GPU
• DeepStream Version 7.0
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs) Question
• 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)

I’m running YoloV8 and trying to increase my detection rate, even at the cost of false negatives. My config is

[property]
gpu-id=0
net-scale-factor=0.0039215697906911373
model-color-format=0
onnx-file=../models/object_detection/yolov8m.onnx
model-engine-file=../models/object_detection/yolov8m_b1_gpu0_fp32.engine
labelfile-path=../models/object_detection/labels.txt
batch-size=1
network-mode=0
num-detected-classes=80
interval=0
gie-unique-id=1
process-mode=1
network-type=0
cluster-mode=2
maintain-aspect-ratio=1
symmetric-padding=1
parse-bbox-func-name=NvDsInferParseYolo
custom-lib-path=../custom_libs/nvdsinfer_custom_impl_Yolo/libnvdsinfer_custom_impl_Yolo.so
engine-create-func-name=NvDsInferYoloCudaEngineGet
filter-out-class-ids=5;9;10;11;12;13;14;15;16;17;18;19;20;21;22;23;24;25;26;27;28;29;30;31;32;33;34;35;36;37;38;39;40;41;42;43;44;45;46;47;48;49;50;51;52;53;54;55;56;57;58;59;60;61;62;63;64;65;66;67;68;69;70;71;72;73;74;75;76;77;78;79;80

[class-attrs-all]
nms-iou-threshold=0.45
pre-cluster-threshold=0.10
post-cluster-threshold=0.10
topk=300

Lowering the pre-cluster-threshold and post-cluster-threshold does not increase the detection rate, however increasing them does lower the detection rate above around 0.6

Expected behavior: Lowering the thresholds increases the detection rate, including false positives
Actual behavior: Lowering the thresholds below 0.3 randomly increases or decreases the number of detections by about 1%.

Is this the intended functionality?

4

it is related to cluster. AYK, cluster algorithm will combine some detection bboxes.

  1. can you try cluster-mode=4? which means " No clustering", then all the bounding box rectangle proposals are returned as it is.
  2. if using cluster-mode=2( NMS), you can set nms-iou-threshold a big value, then most bboxes will be saved.