Types of detection tools for Axxon One

Motion Detection (CPU)

Base motion detection tool when using the СPU resources. Changing the frame rate in the detector settings (the Frames processed per second parameter) does not significantly affect the load. 

Base motion detection tool when using the GPU resources. In this case, the GPU decoder operation mode was used. Changing the frame rate in the detector settings (the Frames processed per second parameter) does not significantly affect the load.

The models and the number of GPUs are selected separately using the information on the GPU performance for Axxon One detection tools page.

Service detection tools for decoding by key frames and use the СPU resources:

• Quality degradation.
• Blurred Image Detection.
• Compression Artifacts Detection.
• Image Noise Detection.
• Scene change.

The platform is calculated for one service detection tool (any of the listed).

The results are given for decoding by key frames if the GOP=25 (every 25th frame is the key frame). The detection tool is applicable only for H.264, H.265 codecs. 

Scene analytics detection tools (VMDA) based on object tracker when using the СPU resources.

The results are given for the object tracker with 1 active sub detection tool Motion in area.

Scene analytics detection tools (VMDA) based on object tracker with use of a neural filter and CPU resources. 

The results are given for a tracker with neural filter with 1 active sub detection tool Motion in area.

Scene analytics detection tools (VMDA) based on object tracker with use of a neural filter and GPU resources. In this case, the CPU decoder operation mode was used.

The results are given for a tracker with neural filter with 1 active sub detection tool Motion in area.

The models and the number of GPUs are selected separately using the information on the GPU performance for Axxon One detection tools page.

Scene analytics detection tools based on neural tracker with use of CPU resources.

The frame rate specified during the Neurotracker object configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The results are given for a standard size neural network^*.

Scene analytics detection tools based on neural tracker with use of VPU resources. In this case, the CPU decoder operation mode was used.

The frame rate specified during the Neurotracker object configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The models and the number of VPUs are selected separately using the information on the VPU performance for Axxon One detection tools page.

The results are given for a standard size neural network^*.

Scene analytics detection tools based on neural tracker with use of GPU resources. In this case, the GPU decoder operation mode was used.

The frame rate specified during the Neurotracker object configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The models and the number of GPUs are selected separately using the information on the GPU performance for Axxon One detection tools page.

The results are given for a standard size neural network^*.

The results are given for a neural tracker with 1 active sub detection tool Motion in area.

Scene analytics detection tools based on neural tracker with use of GPU resources and high-precision neural network. In this case, the GPU decoder operation mode was used.

The frame rate specified during the Neurotracker object configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The models and the number of GPUs are selected separately using the information on the GPU performance for Axxon One detection tools page.

The results are given for a standard size neural network^*.

The results are given for a neural tracker with 1 active sub detection tool Motion in area.

License plate recognition VT detection tool when using the СPU resources.

Face detection tool when using the СPU resources.

Fire detection tool (CPU, 0.1fps)

Smoke detection tool (CPU, 0.1fps)

Fire and smoke detection tools based on neural network with use of CPU resources.

The frame rate specified during the detection tool configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

People counter (CPU)

Pose detection tools based on neural network with use of CPU resources.

The frame rate specified during the detection tool configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The number of specific pose detection tools created under the head Pose detection object does not affect the calculation results (except for the Close-standing people detection; to calculate the result with this detection tool, please contact the AxxonSoft support).

Pose detection tools based on neural network with use of VPU resources. In this case, the CPU decoder operation mode was used.

The frame rate specified during the detection tool configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The number of specific pose detection tools created under the head Pose detection object does not affect the calculation results (except for the Close-standing people detection; to calculate the result with this detection tool, please contact the AxxonSoft support).

The models and the number of VPUs are selected separately using the information on the VPU performance for Axxon One detection tools page.

The results are given for the standard neural network included in the Axxon One distribution.

Equipment detection (CPU, 1fps)

Personal protection equipment (PPE) detection tools based on neural network with use of CPU resources. 

The frame rate specified during the detection tool configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The results are given for a detection tool with 5 classification nets operating simultaneously when determining equipment on each body part (head, torso, hands, legs, feet) in a gateway: at the entrance to the area in which the equipment is required, an employee lingers for 5-10 seconds during which the detection tool determines the presence of the necessary equipment.

Personal protection equipment (PPE) detection tools based on neural network with use of VPU resources. In this case, the CPU decoder operation mode was used.

The frame rate specified during the detection tool configuration (the Frames processed per second parameter) is indicated in brackets. This is the number of fps processed by the module; the frame rate of the incoming video stream is usually higher.

The results are given for a detection tool with 5 classification nets operating simultaneously when determining equipment on each body part (head, torso, hands, legs, feet) in a gateway: at the entrance to the area in which the equipment is required, an employee lingers for 5-10 seconds during which the detection tool determines the presence of the necessary equipment.

The models and the number of VPUs are selected separately using the information on the VPU performance for Axxon One detection tools page.

If you use VPU, please note that due to the peculiarities of the device, only the segmentation neural network will be processed on it, and the CPU will be involved in the operation of the classification neural networks.