SAFR algorithms can operate under widely diverse lighting conditions. SAFR has implemented image-enhancement preprocessing, which dramatically improves video image quality. This capability improves face detection, matching accuracy, and image quality both for automated facial recognition as well as for human review and adjudication. While some high-end IP cameras perform varying image enhancements at the time of image encoding, only SAFR image-enhancement technology has been developed specifically for facial recognition workflows and can be applied universally to any compatible video feed, including the commodity-priced OEM/component cameras typically built into kiosks.
While it may sometimes be obvious, SAFR provides monitoring statistics to determine if the signal from a camera has insufficient lighting. Use the Contrast Quality metric reported in the Recognition Details overlay, which can be activated via the View Menu in the Camera Analyzer window of the SAFR Desktop Client. The Contrast Quality metric should be 0.45 or greater.
If the metric is below 0.45, enable the contrast enhancement feature (located in the Camera Preferences menu in the Desktop Client) to improve recognition. Below is an example of a dark image without contrast enhancement (top) and with contrast enhancement (bottom). Note that the contrast is reported as 0.12 before contrast enhancement is applied and 0.44 after contrast enhancement.