To set up the equipment detection tool (PPE), do the following:

1. To record mask (body-based segmentation) to archive select Yes for the corresponding parameter (1).
   

2. If the camera supports multistreaming, select the stream for which detection is needed (2). 

3. Select one or several files for the classifying neural network (3). There must be a separate classifying neural network to recognize equipment on each body segment. The following classification neural networks for different processor types are located in C:\Program Files\Common Files\AxxonSoft\DetectorPack\NeuroSDK:
   

   ppeHelmet(head)General_movidius.ann	Classification neural network (head) / IntelNCS
   ppeHelmet(head)General_openvino.ann	Classification neural network (head) / CPU
   ppeHelmet(head)General_origin.ann	Classification neural network (head) / GPU
   ppeSafetyVest(body)General_movidius.ann	Classification neural network (body) / IntelNCS
   ppeSafetyVest(body)General_openvino.ann	Classification neural network (body) / CPU
   ppeSafetyVest(body)General_origin.ann	Classification neural network (body) / GPU

   By default, the following are initialized: Classification neural network (head) and Classification neural network (body). When using a unique neural network, you need to specify the path to the file.

   For correct neural network operation under Linux, place the corresponding file in the /opt/AxxonSoft/DetectorPack/NeuroSDK.

   
   

4. Select a processing resource for decoding video streams (4). When you select a GPU, a stand-alone graphics card takes priority (when decoding with NVidia NVDEC chips). If there's no appropriate GPU, the decoding will use the Intel Quick Sync Video technology. Otherwise, CPU resources will be used for decoding.

5. Set the frame rate value for the detection tool to process (5). This value should be in the range [0.016, 100]. 

   To apply detection in gateway mode (see Examples of configuring Equipment detection tool (PPE) for solving typical tasks), we recommend that you use the detection tool standard settings: 1 fps and 3 frames for output (see i.10). To apply detection in continuous mode for busy scenes, set the delay to no less than 4 fps, and the number of frames to no less than 6.

   
   

6. Set the minimum height and width of a person (6, 7) in the frame as a percentage of the frame height/width (0,15 = 15%). Objects which are smaller than the specified size will not be detected.

7. Select the processor for the neural network - CPU, one of GPUs, or Intel processors (8, see Hardware requirements for neural analytics operation).

   It may take several minutes to launch the algorithm on an NVIDIA GPU after you apply the settings. You can use caching to speed up future launches (see Configuring the acceleration of GPU-based neuroanalytics).

   
   

   If you specify other processing resource than the CPU, this device will carry the most of computing load. However, the detection tool will consume CPU as well. If you have Intel HDDL selected, it can host only the segmenting neural networks. The CPU will handle the classifying networks.

   
   

8. Select the segmenting neural network file (9). The following segmenting neural networks for different processor types are located in C:\Program Files\Common Files\AxxonSoft\DetectorPack\NeuroSDK:
   

   ppeSegmentationByPose_movidius.ann	Segmenting neural network (head, body) / IntelNCS
   ppeSegmentationByPose_openvino.ann	Segmenting neural network (head, body) / CPU
   ppeSegmentationByPose_origin_onnx.ann	Segmenting neural network (head, body) / GPU

   By default, the Segmenting neural network (head, body) is initialized. When using a unique neural network, you need to specify the path to the file.
   

   For correct neural network operation under Linux, place the corresponding file in the /opt/AxxonSoft/DetectorPack/NeuroSDK.

   
   

9. Select the Mask checkbox to display body segments in the preview window (10).
   

10. Set the minimum number of frames containing people with no PPE for triggering the tool - Number of measurements in a row to trigger detection (11). The value should be within the range of 2 – 20.

11. By default, each equipment element's triggering occurs once during a continuous tracking of a human object.  You can set triggering to multiple by setting the One Event per Equipment Element parameter to No (12).

    Example. An individual not wearing a helmet appears in FoV, puts on a helmet, then puts it off. If the One Event per Equipment Element parameter is activated, you will have one alarm event, otherwise two.

    
    

12. In the preview window, you can set the detection zones with the help of anchor points much like privacy masks in Scene Analytics (see Setting General Zones for Scene Analytics). By default, the entire FoV is a detection zone.
13. Click Apply.

Equipment detection tool (PPE) is now configured. 

The equipment detection tool (PPE) triggers an alarm when a person not wearing required equipment (PPE) on specified body parts, or wearing inappropriate equipment, appears in FoV.

Equipment detection tool (PPE) recognizes equipment of the following colors:

1. Helmets:
   1. Yellow.
   2. White.
   3. Blue.
   4. Green.
   5. Orange.
   6. Black.
   7. Red.
      
2. Vests:
   1. Yellow.
   2. Orange.