How easy is it to detect different personal protective equipment (PPE) ?

Personal Protective Equipment (PPE) is highly diverse, affecting the reliability of detection, as different shapes and sizes of the same item can differ according to specific PPE types and situational factors. Classifying the presence or absence of certain PPEs reliably depends on two aspects: the difficulty of detecting the PPE itself, and the plausibility of classifying the detected PPE correctly.

Difficulty and reliability of correct detections depend on two main factors

PPE that is easy to detect and has high classification plausibility includes helmets, masks, sleeves, and other PPEs with highly consistent and distinct shapes, as the perception of their shapes does not change due to different camera angles or distortion. This leads to PPE classification with higher reliability, allowing us to expand solutions to detect company specific PPE. For example, rather than detecting all types of helmets, we can distinguish between specific helmets based on their color or shape.

Examples of PPE that are harder to detect, but have high classification plausibility are safety glasses and earmuffs. The transparent color of safety glasses makes them difficult to detect, especially at longer distances. However, once detected, it is easier to classify glasses as safety glasses since they possess distinctive features such as lense and frame shape.

Plausibility of detection & plausibility of classification of different PPE

A special example are protective boots, which are easier to detect, but they have low plausibility of classification. Since they lack any distinguishing markings, it is hard to classify boots as safety boots. Additional problems arise from the boot design, as most safety boots are designed to look similar to normal boots or shoes, thus lowering the reliability of detection.

Although the type of PPE does affect the level of reliability of detection, our solutions maintain the utmost reliability, as we enhance their situational understanding by using multiple cameras simultaneously. This allows us to pivot through corner cases and mitigate the risks of noisy sensors. Additionally, our solutions’ reliability of detection is not impacted by lighting conditions, as we use novel methods for data engineering.

Get in touch to learn more about PPE detection solutions and our Bleenco Go platform.