real time analytics platform,

Identification, Landmarking

Recognize faces inside an edge, create key focuses on the face, and start following stay time and session time. Pass that location into our neural systems for our socioeconomics.

Neural Network

When we recognize and milestone, the recognition is passed to our neural system to produce a chain of importance of highlights, to contrast with our current model of what our socioeconomics should resemble.

Produce Demographics

From that point, we create age, sexual orientation, and feeling for identified individual and push these measurements to the dashboard, trigger webhooks, and store in our database.

Exactness and Metrics

The following are refreshed factual examination of the precision of our information focuses. We enhance these every day and will develop the sorts of measurements we assemble as we further build up our innovation stage.

Metrics Accuracy (%)

Happy 80.5

Angry 83.3

Surprised 91.1

Fear 82.6

Sad 78.1

Disgust 98.5

Calm 80.8

Metrics Accuracy (%)

Male 91.3

Female 91.3

Age +/ - 10 years

http://muzun.ae/technology.html

View +/ - 5 degrees (91.3%)

Stay Time +/ - 100 ms

Uniqueness and Occlusion Algorithms

We utilize a mix of complex following and recognition calculations to keep up an individual's session over the screen.

Our intricate calculations can deal with impediments, hindrances, and lost location, to keep up the uniqueness of an individual over the session.

A client's "session" is considered from the time the face is first recognized, till an interim of time that the face can never again be identified.

Low Latency

Our Computer Vision calculations consider low idleness executions, on minimal effort equipment arrangements on location, or remotely.

We've improved our discoveries to process faces as low as 2 milliseconds for each face at a separation of roughly 25-30 feet.

Range

Our Computer Vision innovation can be connected to any number of cameras with short and long-extend vision.

The nature of our measurements are autonomous of the range, and absolutely dependent on the pixel measurements of the identifications.

Our testing reach and least location measure is 30x30 pixels, at a separation of 25 feet, and can be stretched out with higher goals cameras.

Multi-Platform Support

Our libraries were intended to be effortlessly deployable and bolster both OpenCV and OpenCL reconciliations.

On the best way to begin with our SDKs, if it's not too much trouble get in touch with us.