The PPE detection tool locates people wearing no personal protective equipment within the area where it's required.
For detection tool's operation, you need at least two separate neural networks:
- a segmenting network structures up an image of a human body (locates head, shoulders, arms, hands, thighs, legs and feet);
- a classifying network detects PPE on a specified body part.
A PPE detection tool may use:
- its own metadata — in this case, the detection tool must be created under a camera parent object;
- metadata from the neural tracker — this requires detection tool creation under the neural tracker object.
Applying neural tracker may level up detection quality for the cost of increasing CPU load. |