Recently, facial recognition technology has advanced
significantly and become widely used. However, detecting faces from
surveillance camera footage in real time can be a difficult task due to
a variety of factors. One major challenge is that the quality of the
footage may be poor, with low resolution or high levels of noise.
Additionally, lighting conditions can also make it difficult to detect
faces, especially in low light environments. Furthermore, faces may be
partially obscured by objects such as hats or sunglasses, or by changes
in pose or expression. Finally, the number of faces in a given scene can
also make it difficult to detect and track individuals in real-time.
time is the presence of occlusions. Occlusions occur when an object or
person blocks the view of a face, making it difficult or impossible to
detect. Additionally, faces may be captured at different angles and
scales, which can further complicate the detection process. Another
factor that makes face detection difficult is that faces can appear
differently due to various factors such as age, gender, race, and facial
expressions. This can make it hard for the algorithm to generalize and
detect all types of faces. In this article, we propose GetFace, an
intelligent system for enhancing surveillance footage. GetFace uses
CNNs, super-resolution methods, human shape and face detection
techniques, Wiener filter, and face clustering to improve the image
quality.