Camera requirements for the operation of the PPE detection VL are given in the table:
Camera
Resolution must be at least 640x360. We also don't recommend using a resolution higher than 1920×1080, since higher resolution doesn't increase the detection quality, but significantly increases the consumption of resources. The optimal resolution for solving typical tasks is 1920x1080
The frame rate per second in the video stream from the camera is at least 13 for solving typical tasks
Detection-based analytics works with color images only
Lighting
Lighting in the scene is at least 200 lux. The detection tool is applicable in conditions of artificial light. In conditions of insufficient or excessive lighting (night or light-striking), stable operation of the analytics isn't guaranteed
No abrupt changes in lighting
Scene and camera angle
Moving objects are visually separable from each other
The background is mostly static and doesn't change abruptly
There are no products made of rods in the detection area
Moving objects are minimally obscured by static objects in the scene (columns, industrial products, and so on)
Avoid large obstructions in the camera FOV (columns, tower cranes, stacked materials, and so on). They will obscure the detected objects and cause track breaks. If obstructions cannot be avoided, we recommend not placing them in the center of the frame
The analyzed scene doesn't have reflective surfaces. If present, they must be masked
The analyzed scene doesn't have strong light flashes (spotlights, active welding zones)
Camera shake doesn't result in image offsets greater than 1% of the frame size
Shooting angle without a strong camera tilt
To improve the quality of the video stream, the camera lens must be cleaned of heavy dirt. We don't recommend locating cameras above the material unloading area or near ventilation shafts
Object images
The detected object (equipment or PPE) must be clearly distinguishable by the human eye
Color image, isn't noisy and isn't distorted by the artifacts of compression algorithm
Linear dimensions of equipment or PPE must not exceed 65% of the corresponding frame size
The duration of the object's visibility is at least 10–13 frames. The minimum number of frames depends on the task
There are no visible physical barriers between the camera lens and the analyzed object
Obscuring of a person and/or PPE by an obstruction and/or by another person, and/or cropping by frame boundaries must not exceed 25%
The minimum sizes of equipment on body parts in pixels are observed:
Vest—at least 50 pixels
Helmet/Hood—at least 20 pixels
PPE must have a standard look and shape, and have the following colors:
Vest: green, yellow, orange
Helmet/Hood: white, yellow, orange, red, blue, green
The minimum value of pixel density per meter is observed:
Image resolution
Minimum pixel density per meter (ratio of the object height in pixels to the object height in meters)
Minimum object height in pixels
1920x1080
60
100
1280x720
60
100
640x360
60
100
Correct operation of the detection tool isn't guaranteed when using a fisheye lens.